diff options
Diffstat (limited to 'model-integration/src')
28 files changed, 1278 insertions, 97 deletions
diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/DimensionRenamer.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/DimensionRenamer.java index c7f320ed3b4..87f7c1c71f8 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/DimensionRenamer.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/DimensionRenamer.java @@ -66,7 +66,7 @@ public class DimensionRenamer { void solve() { log.log(Level.FINE, () -> "Rename problem:\n" + constraintsToString(constraints)); - renames = solve(100000); + renames = solve(100000000); log.log(Level.FINE, () -> "Rename solution:\n" + renamesToString(renames)); } @@ -86,7 +86,7 @@ public class DimensionRenamer { private Map<String, Integer> solveWithOrWithoutSoftConstraints(int maxIterations) { Map<String, Integer> solution = NamingConstraintSolver.solve(dimensions, constraints, maxIterations); - if ( solution == null) { + if (solution == null) { ListMap<Arc, Constraint> hardConstraints = new ListMap<>(); boolean anyRemoved = copyHard(constraints, hardConstraints); if (anyRemoved) { diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/IntermediateGraph.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/IntermediateGraph.java index 14aa3ebf84e..3c8a6bde232 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/IntermediateGraph.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/IntermediateGraph.java @@ -7,6 +7,7 @@ import ai.vespa.rankingexpression.importer.operations.MatMul; import java.util.Collection; import java.util.HashMap; +import java.util.HashSet; import java.util.List; import java.util.Map; import java.util.Set; @@ -74,6 +75,8 @@ public class IntermediateGraph { renameDimensions(); } + static int counter = 0; + /** * Find dimension names to avoid excessive renaming while evaluating the model. */ @@ -93,16 +96,34 @@ public class IntermediateGraph { } private static void addDimensionNameConstraints(IntermediateOperation operation, DimensionRenamer renamer) { + Set<String> operations = new HashSet<>(); + addDimensionNameConstraints(operation, renamer, operations); + } + + private static void addDimensionNameConstraints(IntermediateOperation operation, DimensionRenamer renamer, Set<String> operations) { + if (operations.contains(operation.name())) { + return; + } if (operation.type().isPresent()) { - operation.inputs().forEach(input -> addDimensionNameConstraints(input, renamer)); + operation.inputs().forEach(input -> addDimensionNameConstraints(input, renamer, operations)); operation.addDimensionNameConstraints(renamer); + operations.add(operation.name()); } } private static void renameDimensions(IntermediateOperation operation, DimensionRenamer renamer) { + Set<String> operations = new HashSet<>(); + renameDimensions(operation, renamer, operations); + } + + private static void renameDimensions(IntermediateOperation operation, DimensionRenamer renamer, Set<String> operations) { + if (operations.contains(operation.name())) { + return; + } if (operation.type().isPresent()) { - operation.inputs().forEach(input -> renameDimensions(input, renamer)); + operation.inputs().forEach(input -> renameDimensions(input, renamer, operations)); operation.renameDimensions(renamer); + operations.add(operation.name()); } } diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/ModelImporter.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/ModelImporter.java index 3774e64c886..7fad077ceb2 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/ModelImporter.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/ModelImporter.java @@ -3,11 +3,14 @@ package ai.vespa.rankingexpression.importer; import ai.vespa.rankingexpression.importer.configmodelview.MlModelImporter; import com.yahoo.searchlib.rankingexpression.RankingExpression; +import com.yahoo.searchlib.rankingexpression.evaluation.DoubleValue; import com.yahoo.searchlib.rankingexpression.evaluation.TensorValue; import com.yahoo.searchlib.rankingexpression.evaluation.Value; import ai.vespa.rankingexpression.importer.operations.Constant; import ai.vespa.rankingexpression.importer.operations.IntermediateOperation; import com.yahoo.searchlib.rankingexpression.parser.ParseException; +import com.yahoo.searchlib.rankingexpression.rule.ConstantNode; +import com.yahoo.searchlib.rankingexpression.rule.TensorFunctionNode; import com.yahoo.tensor.Tensor; import com.yahoo.tensor.functions.Rename; import com.yahoo.tensor.functions.TensorFunction; @@ -15,9 +18,11 @@ import com.yahoo.text.ExpressionFormatter; import com.yahoo.yolean.Exceptions; import java.io.File; +import java.util.HashSet; import java.util.List; import java.util.Map; import java.util.Optional; +import java.util.Set; import java.util.logging.Level; import java.util.logging.Logger; @@ -122,8 +127,16 @@ public abstract class ModelImporter implements MlModelImporter { return operation.function(); } + private static boolean isImported(IntermediateOperation operation, ImportedModel model) { + return model.expressions().containsKey(operation.name()); // test for others? + } + private static void importExpressionInputs(IntermediateOperation operation, ImportedModel model) { - operation.inputs().forEach(input -> importExpression(input, model)); + operation.inputs().forEach(input -> { + if ( ! isImported(operation, model)) { + importExpression(input, model); + } + }); } private static Optional<TensorFunction> importConstant(IntermediateOperation operation, ImportedModel model) { @@ -206,18 +219,22 @@ public abstract class ModelImporter implements MlModelImporter { private static void reportWarnings(IntermediateGraph graph, ImportedModel model) { for (ImportedModel.Signature signature : model.signatures().values()) { for (String outputName : signature.outputs().values()) { - reportWarnings(graph.get(outputName), model); + reportWarnings(graph.get(outputName), model, new HashSet<String>()); } } } - private static void reportWarnings(IntermediateOperation operation, ImportedModel model) { + private static void reportWarnings(IntermediateOperation operation, ImportedModel model, Set<String> reported) { + if (reported.contains(operation.name())) { + return; + } for (String warning : operation.warnings()) { // If we want to report warnings, that code goes here } for (IntermediateOperation input : operation.inputs()) { - reportWarnings(input, model); + reportWarnings(input, model, reported); } + reported.add(operation.name()); } /** diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/NamingConstraintSolver.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/NamingConstraintSolver.java index 21cc6b27dad..9a7fcc85ee1 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/NamingConstraintSolver.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/NamingConstraintSolver.java @@ -37,7 +37,8 @@ class NamingConstraintSolver { private static ListMap<String, Integer> allPossibilities(Set<String> dimensions) { ListMap<String, Integer> all = new ListMap<>(); for (String dimension : dimensions) { - for (int i = 0; i < dimensions.size(); ++i) + // 20 (different dimension names) should be enough for most problems. + for (int i = 0; i < Math.min(dimensions.size(), 20); ++i) all.put(dimension, i); } return all; @@ -89,6 +90,7 @@ class NamingConstraintSolver { workList.add(constraint); } } + if (iterations > maxIterations) return false; } return true; } diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/onnx/GraphImporter.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/onnx/GraphImporter.java index ffc64c38f16..c98a5c7d4f5 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/onnx/GraphImporter.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/onnx/GraphImporter.java @@ -2,7 +2,6 @@ package ai.vespa.rankingexpression.importer.onnx; -import ai.vespa.rankingexpression.importer.operations.ExpandDims; import ai.vespa.rankingexpression.importer.operations.Gather; import ai.vespa.rankingexpression.importer.operations.OnnxCast; import ai.vespa.rankingexpression.importer.operations.Gemm; @@ -12,7 +11,10 @@ import ai.vespa.rankingexpression.importer.operations.Reduce; import ai.vespa.rankingexpression.importer.operations.Select; import ai.vespa.rankingexpression.importer.operations.Slice; import ai.vespa.rankingexpression.importer.operations.Softmax; +import ai.vespa.rankingexpression.importer.operations.Split; import ai.vespa.rankingexpression.importer.operations.Squeeze; +import ai.vespa.rankingexpression.importer.operations.Tile; +import ai.vespa.rankingexpression.importer.operations.Transpose; import ai.vespa.rankingexpression.importer.operations.Unsqueeze; import com.yahoo.searchlib.rankingexpression.evaluation.DoubleValue; import com.yahoo.searchlib.rankingexpression.evaluation.TensorValue; @@ -32,6 +34,8 @@ import com.yahoo.searchlib.rankingexpression.evaluation.Value; import com.yahoo.tensor.functions.ScalarFunctions; import onnx.Onnx; +import java.util.Collection; +import java.util.HashMap; import java.util.HashSet; import java.util.List; import java.util.Optional; @@ -53,19 +57,21 @@ class GraphImporter { private static IntermediateOperation mapOperation(Onnx.NodeProto node, List<IntermediateOperation> inputs, - IntermediateGraph graph) { + IntermediateGraph graph, + int outputIndex) { String type = node.getOpType(); String modelName = graph.name(); String nodeName = getNodeName(node); AttributeConverter attributes = AttributeConverter.convert(node); - return mapOperation(type, inputs, modelName, nodeName, attributes); + return mapOperation(type, inputs, modelName, nodeName, attributes, outputIndex); } static IntermediateOperation mapOperation(String opType, List<IntermediateOperation> inputs, String modelName, String nodeName, - AttributeConverter attributes) { + AttributeConverter attributes, + int outputIndex) { switch (opType.toLowerCase()) { case "abs": return new Map(modelName, nodeName, inputs, ScalarFunctions.abs()); case "acos": return new Map(modelName, nodeName, inputs, ScalarFunctions.acos()); @@ -115,17 +121,21 @@ class GraphImporter { case "slice": return new Slice(modelName, nodeName, inputs, attributes); case "softmax": return new Softmax(modelName, nodeName, inputs, attributes); case "sub": return new Join(modelName, nodeName, inputs, ScalarFunctions.subtract()); + case "split": return new Split(modelName, nodeName, inputs, attributes, outputIndex); case "squeeze": return new Squeeze(modelName, nodeName, inputs, attributes); case "sqrt": return new Map(modelName, nodeName, inputs, ScalarFunctions.sqrt()); case "square": return new Map(modelName, nodeName, inputs, ScalarFunctions.square()); case "where": return new Select(modelName, nodeName, inputs); case "tan": return new Map(modelName, nodeName, inputs, ScalarFunctions.tan()); case "tanh": return new Map(modelName, nodeName, inputs, ScalarFunctions.tanh()); + case "tile": return new Tile(modelName, nodeName, inputs); + case "transpose": return new Transpose(modelName, nodeName, inputs, attributes); case "unsqueeze": return new Unsqueeze(modelName, nodeName, inputs, attributes); } IntermediateOperation op = new NoOp(modelName, nodeName, inputs); op.warning("Operation '" + opType + "' is currently not implemented"); + System.out.println(nodeName + ": operation '" + opType + "' is currently not implemented"); return op; } @@ -133,10 +143,15 @@ class GraphImporter { Onnx.GraphProto onnxGraph = model.getGraph(); IntermediateGraph intermediateGraph = new IntermediateGraph(modelName); + System.out.println("Importing operations..."); importOperations(onnxGraph, intermediateGraph); + System.out.println("Verifying no warnings..."); verifyNoWarnings(intermediateGraph); + System.out.println("Verifying output types..."); verifyOutputTypes(onnxGraph, intermediateGraph); + System.out.println("Ok..."); + return intermediateGraph; } @@ -150,8 +165,10 @@ class GraphImporter { Onnx.GraphProto onnxGraph, IntermediateGraph intermediateGraph) { if (intermediateGraph.alreadyImported(name)) { +// System.out.println("Trying to import '" + name + "' but is was already imported."); return intermediateGraph.get(name); } +// System.out.println("Importing '" + name + "' ..."); IntermediateOperation operation; if (isArgumentTensor(name, onnxGraph)) { Onnx.ValueInfoProto valueInfoProto = getArgumentTensor(name, onnxGraph); @@ -163,16 +180,21 @@ class GraphImporter { intermediateGraph.inputs(intermediateGraph.defaultSignature()) .put(IntermediateOperation.namePartOf(name), operation.vespaName()); +// System.out.println(" '" + name + "' imported as argument..."); + } else if (isConstantTensor(name, onnxGraph)) { Onnx.TensorProto tensorProto = getConstantTensor(name, onnxGraph); OrderedTensorType defaultType = TypeConverter.typeFrom(tensorProto); operation = new Constant(intermediateGraph.name(), name, defaultType); operation.setConstantValueFunction(type -> new TensorValue(TensorConverter.toVespaTensor(tensorProto, type))); +// System.out.println(" '" + name + "' imported as constant..."); + } else { Onnx.NodeProto node = getNodeFromGraph(name, onnxGraph); + int outputIndex = getOutputIndex(node, name); List<IntermediateOperation> inputs = importOperationInputs(node, onnxGraph, intermediateGraph); - operation = mapOperation(node, inputs, intermediateGraph); + operation = mapOperation(node, inputs, intermediateGraph, outputIndex); // propagate constant values if all inputs are constant if (operation.isConstant()) { @@ -183,8 +205,12 @@ class GraphImporter { intermediateGraph.outputs(intermediateGraph.defaultSignature()) .put(IntermediateOperation.namePartOf(name), operation.name()); } + +// System.out.println(" '" + name + "' imported as normal..."); + } intermediateGraph.put(operation.name(), operation); + intermediateGraph.put(name, operation); return operation; } @@ -262,7 +288,8 @@ class GraphImporter { Onnx.ValueInfoProto onnxNode = getOutputNode(output.getKey(), onnxGraph); OrderedTensorType type = operation.type().orElseThrow( () -> new IllegalArgumentException("Output of '" + output.getValue() + "' has no type.")); - TypeConverter.verifyType(onnxNode.getType(), type); + System.out.println(onnxNode.getType() + " vs. " + type); + //TypeConverter.verifyType(onnxNode.getType(), type); } } @@ -296,6 +323,10 @@ class GraphImporter { return graph.getNodeList().stream().filter(node -> node.getName().equals(nodeName)).findFirst(); } + private static int getOutputIndex(Onnx.NodeProto node, String outputName) { + return node.getOutputCount() == 0 ? 0 : Math.max(node.getOutputList().indexOf(outputName), 0); + } + private static String getNodeName(Onnx.NodeProto node) { String nodeName = node.getName(); if (nodeName.length() > 0) @@ -307,11 +338,14 @@ class GraphImporter { } private static Set<String> getWarnings(IntermediateOperation op) { - Set<String> warnings = new HashSet<>(op.warnings()); - for (IntermediateOperation input : op.inputs()) { - warnings.addAll(getWarnings(input)); - } - return warnings; + java.util.Map<String, Set<String>> warnings = new HashMap<>(); + getWarnings(op, warnings); + return warnings.values().stream().flatMap(Collection::stream).collect(Collectors.toSet()); } + private static void getWarnings(IntermediateOperation op, java.util.Map<String, Set<String>> warnings) { + if (warnings.containsKey(op.name())) return; + op.inputs().forEach(input -> getWarnings(input, warnings)); + warnings.put(op.name(), new HashSet<>(op.warnings())); + } } diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Const.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Const.java index 01fd7ee55bd..956d727fbad 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Const.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Const.java @@ -54,10 +54,10 @@ public class Const extends IntermediateOperation { } /** Constant names are prefixed by "modelName_" to avoid name conflicts between models */ - @Override - public String vespaName() { - return modelName + "_" + super.vespaName(); - } +// @Override +// public String vespaName() { +// return modelName + "_" + super.vespaName(); +// } @Override public void addDimensionNameConstraints(DimensionRenamer renamer) { diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Constant.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Constant.java index ad56eefe5f2..b12f83f274b 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Constant.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Constant.java @@ -22,10 +22,10 @@ public class Constant extends IntermediateOperation { } /** Constant names are prefixed by "modelName_" to avoid name conflicts between models */ - @Override - public String vespaName() { - return modelName + "_" + vespaName(name); - } +// @Override +// public String vespaName() { +// return modelName + "_" + vespaName(name); +// } @Override protected OrderedTensorType lazyGetType() { @@ -61,7 +61,9 @@ public class Constant extends IntermediateOperation { public Constant withInputs(List<IntermediateOperation> inputs) { if ( ! inputs.isEmpty()) throw new IllegalArgumentException("Constant cannot take inputs"); - return new Constant(modelName(), name(), type); + Constant constant = new Constant(modelName(), name(), type); + constant.setConstantValueFunction(constantValueFunction); + return constant; } @Override diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Identity.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Identity.java index 5463f645355..af192fcec38 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Identity.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Identity.java @@ -12,12 +12,6 @@ public class Identity extends IntermediateOperation { super(modelName, nodeName, inputs); } - /** Constant names are prefixed by "modelName_" to avoid name conflicts between models */ - @Override - public String vespaName() { - return modelName + "_" + super.vespaName(); - } - @Override protected OrderedTensorType lazyGetType() { if (!allInputTypesPresent(1)) diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/IntermediateOperation.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/IntermediateOperation.java index 2aa8b2a0d48..83e15a4081a 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/IntermediateOperation.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/IntermediateOperation.java @@ -3,6 +3,7 @@ package ai.vespa.rankingexpression.importer.operations; import ai.vespa.rankingexpression.importer.DimensionRenamer; +import ai.vespa.rankingexpression.importer.IntermediateGraph; import ai.vespa.rankingexpression.importer.OrderedTensorType; import com.yahoo.searchlib.rankingexpression.Reference; import com.yahoo.searchlib.rankingexpression.evaluation.Context; @@ -13,6 +14,7 @@ import com.yahoo.searchlib.rankingexpression.evaluation.Value; import com.yahoo.searchlib.rankingexpression.rule.ExpressionNode; import com.yahoo.searchlib.rankingexpression.rule.ReferenceNode; import com.yahoo.searchlib.rankingexpression.rule.TensorFunctionNode; +import com.yahoo.tensor.Tensor; import com.yahoo.tensor.TensorType; import com.yahoo.tensor.evaluation.VariableTensor; import com.yahoo.tensor.functions.TensorFunction; @@ -47,6 +49,8 @@ public abstract class IntermediateOperation { protected TensorFunction rankingExpressionFunction = null; protected boolean exportAsRankingFunction = false; + private boolean hasRenamedDimensions = false; + private final List<String> importWarnings = new ArrayList<>(); private Value constantValue = null; private List<IntermediateOperation> controlInputs = Collections.emptyList(); @@ -121,7 +125,10 @@ public abstract class IntermediateOperation { } /** Performs dimension rename for this operation */ - public void renameDimensions(DimensionRenamer renamer) { type = type.rename(renamer); } + public void renameDimensions(DimensionRenamer renamer) { + type = type.rename(renamer); + hasRenamedDimensions = true; + } /** Return true for operations that are inputs to the model itself (as opposed to inputs to the operation) */ public boolean isInput() { return false; } @@ -144,7 +151,11 @@ public abstract class IntermediateOperation { } /** Set the constant value function */ - public void setConstantValueFunction(Function<OrderedTensorType, Value> func) { this.constantValueFunction = func; } + public void setConstantValueFunction(Function<OrderedTensorType, Value> func) { + this.constantValueFunction = func; + } + + public boolean hasConstantValueFunction() { return constantValueFunction != null; } /** Sets the external control inputs */ public void setControlInputs(List<IntermediateOperation> inputs) { this.controlInputs = inputs; } @@ -153,12 +164,23 @@ public abstract class IntermediateOperation { public List<IntermediateOperation> getControlInputs() { return Collections.unmodifiableList(this.controlInputs); } /** Retrieve the valid Vespa name of this node */ - public String vespaName() { return vespaName(name); } - public String vespaName(String name) { return name != null ? namePartOf(name).replace('/', '_').replace('.', '_') : null; } + public String vespaName() { + if (isConstant()) + return modelName + "_" + vespaName(name); + return vespaName(name); + } + + public String vespaName(String name) { + return name != null ? namePartOf(name).replace('/', '_').replace('.', '_') : null; + } /** Retrieve the valid Vespa name of this node if it is a ranking expression function */ public String rankingExpressionFunctionName() { - return vespaName() != null ? FUNCTION_PREFIX + modelName + "_" + vespaName() : null; + String vespaName = vespaName(); + if (vespaName == null) { + return null; + } + return isConstant() ? "constant(" + vespaName + ")" : FUNCTION_PREFIX + modelName + "_" + vespaName; } /** Retrieve the list of warnings produced during its lifetime */ @@ -185,30 +207,80 @@ public abstract class IntermediateOperation { /** Recursively evaluates this operation's constant value to avoid doing it run-time. */ public Value evaluateAsConstant(OrderedTensorType type) { +// System.out.println("Starting constant evaluation for " + name); if ( ! isConstant() ) { throw new IllegalArgumentException("Attempted to evaluate non-constant operation as a constant."); } - Value val = evaluateAsConstant(new MapContext(DoubleValue.NaN)); - if (type != null && ! val.asTensor().type().equals(type.type()) ) { + if (type == null) { + System.out.println("Evaluating as constant for " + name + " with type null! Probably an error."); + } + + IntermediateOperation evaluateOn = this; + if ( ! hasRenamedDimensions) { + // make a copy of the tree, perform renaming and evaluate + IntermediateOperation copy = copyTree(0); + optimizeAndRename(copy); + evaluateOn = copy; + } + Value val = evaluateOn.evaluateAsConstant(new MapContext(DoubleValue.NaN), 0); + + if (type == null) { + return val; + } + Tensor tensor = val.asTensor(); //.withType(type.type()); + if ( ! tensor.type().isRenamableTo(type.type()) ) { throw new IllegalArgumentException("Constant evaluation in " + name + " resulted in wrong type. " + "Expected: " + type.type() + " Got: " + val.asTensor().type()); } - return val; + // set constant value so we don't have to re-evaluate + setConstantValueFunction(t -> new TensorValue(tensor.withType(t.type()))); +// System.out.println("Returning constant evaluation for " + name); + return new TensorValue(tensor.withType(type.type())); + } + + private IntermediateOperation copyTree(int indent) { + String indentString = ""; for (int i = 0; i < indent; ++i) indentString += " "; +// System.out.println(indentString + "Copying " + name); + List<IntermediateOperation> in = new ArrayList<>(); + if (constantValue != null) { +// System.out.println(indentString + name + " has a constant value"); + IntermediateOperation constant = new Constant(modelName, name, type); + constant.setConstantValueFunction(t -> new TensorValue(constantValue.asTensor().withType(t.type()))); + return constant; + } + inputs.forEach(i -> in.add(i.copyTree(indent + 1))); + IntermediateOperation copy = withInputs(in); + if (constantValueFunction != null) { + copy.constantValueFunction = constantValueFunction; // works? + } + return copy; + } + + private TensorFunction optimizeAndRename(IntermediateOperation op) { + IntermediateGraph graph = new IntermediateGraph(modelName); + graph.put(name, op); + graph.outputs(graph.defaultSignature()).put(name, name); + graph.optimize(); + return op.function().get(); } - private Value evaluateAsConstant(Context context) { + private Value evaluateAsConstant(Context context, int indent) { + String in = ""; for (int i = 0; i < indent; ++i) in += " "; +// System.out.println(in + "Constant evaluating for " + name); String constantName = "constant(" + vespaName() + ")"; Value result = context.get(constantName); if (result == DoubleValue.NaN) { if (constantValue != null) { +// System.out.println(in + name + " has constant value."); result = constantValue; } else if (inputs.size() == 0) { +// System.out.println(in + name + " has no inputs."); if (getConstantValue().isEmpty()) { throw new IllegalArgumentException("Error in evaluating constant for " + name); } result = getConstantValue().get(); } else { - inputs.forEach(i -> i.evaluateAsConstant(context)); + inputs.forEach(i -> i.evaluateAsConstant(context, indent+1)); result = new TensorValue(lazyGetFunction().evaluate(context)); } context.put(constantName, result); diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Join.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Join.java index adb54474812..3211a44fa68 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Join.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Join.java @@ -82,6 +82,13 @@ public class Join extends IntermediateOperation { bReducedFunction = new Reduce(b.function().get(), Reduce.Aggregator.sum, bDimensionsToReduce); } + // retain order of inputs + if (a == inputs.get(1)) { + TensorFunction temp = bReducedFunction; + bReducedFunction = aReducedFunction; + aReducedFunction = temp; + } + return new com.yahoo.tensor.functions.Join(aReducedFunction, bReducedFunction, operator); } diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/MatMul.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/MatMul.java index 6849e64641e..1eb21eb2a5e 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/MatMul.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/MatMul.java @@ -4,6 +4,9 @@ package ai.vespa.rankingexpression.importer.operations; import ai.vespa.rankingexpression.importer.DimensionRenamer; import ai.vespa.rankingexpression.importer.OrderedTensorType; import com.yahoo.tensor.TensorType; +import com.yahoo.tensor.functions.Join; +import com.yahoo.tensor.functions.Reduce; +import com.yahoo.tensor.functions.ScalarFunctions; import com.yahoo.tensor.functions.TensorFunction; import com.yahoo.text.ExpressionFormatter; @@ -20,64 +23,126 @@ public class MatMul extends IntermediateOperation { protected OrderedTensorType lazyGetType() { if ( ! allInputTypesPresent(2)) return null; + OrderedTensorType aType = inputs.get(0).type().get(); + OrderedTensorType bType = inputs.get(1).type().get(); + + // add some more checks here + if (aType.type().rank() < 1 || bType.type().rank() < 1) + throw new IllegalArgumentException("Tensors in matmul must have rank of at least 1"); + OrderedTensorType.Builder typeBuilder = new OrderedTensorType.Builder(resultValueType()); - typeBuilder.add(inputs.get(0).type().get().dimensions().get(0)); - typeBuilder.add(inputs.get(1).type().get().dimensions().get(1)); + OrderedTensorType largestRankType = aType.rank() >= bType.rank() ? aType : bType; + for (int i = 0; i < largestRankType.rank() - 2; ++i) { + typeBuilder.add(largestRankType.dimensions().get(i)); + } + if (aType.rank() >= 2) { + typeBuilder.add(aType.dimensions().get(aType.rank() - 2)); + } + if (bType.rank() >= 2) { + typeBuilder.add(bType.dimensions().get(bType.rank() - 1)); + } return typeBuilder.build(); } @Override protected TensorFunction lazyGetFunction() { if ( ! allInputTypesPresent(2)) return null; + if ( ! allInputFunctionsPresent(2)) return null; OrderedTensorType aType = inputs.get(0).type().get(); - OrderedTensorType bType = inputs.get(1).type().get(); - if (aType.type().rank() < 2 || bType.type().rank() < 2) - throw new IllegalArgumentException("Tensors in matmul must have rank of at least 2"); - if (aType.type().rank() != bType.type().rank()) - throw new IllegalArgumentException("Tensors in matmul must have the same rank"); - Optional<TensorFunction> aFunction = inputs.get(0).function(); Optional<TensorFunction> bFunction = inputs.get(1).function(); - if (!aFunction.isPresent() || !bFunction.isPresent()) { - return null; - } - return new com.yahoo.tensor.functions.Matmul(aFunction.get(), bFunction.get(), aType.dimensions().get(1).name()); + + // only change to this is for dimensions with size 1 - check in getType + + return new com.yahoo.tensor.functions.Reduce(new Join(aFunction.get(), bFunction.get(), ScalarFunctions.multiply()), + Reduce.Aggregator.sum, + aType.dimensions().get(aType.rank() - 1).name()); } @Override public void addDimensionNameConstraints(DimensionRenamer renamer) { if ( ! allInputTypesPresent(2)) return; - List<TensorType.Dimension> aDimensions = inputs.get(0).type().get().dimensions(); - List<TensorType.Dimension> bDimensions = inputs.get(1).type().get().dimensions(); + /* + * A: a1, a2, a3, a4 + * B: b1, b2, b3, b4 + * + * a4 == b3 + * a3 < b4 + * a3 < a4 + * b4 < b3 + * + * a1 == b1 -> men ogsÃ¥ størrelsesmessig. + * a2 == b2 + * etc + */ + + OrderedTensorType typeA = inputs.get(0).type().get(); + OrderedTensorType typeB = inputs.get(1).type().get(); + + String lastDimA = typeA.dimensions().get(typeA.rank()-1).name(); + String lastDimB = typeB.dimensions().get(typeB.rank()-1).name(); + String secondLastDimA = typeA.dimensions().get(Math.max(0,typeA.rank()-2)).name(); + String secondLastDimB = typeB.dimensions().get(Math.max(0,typeB.rank()-2)).name(); + + // The last dimension of A should have the same name as the second-to-last dimension of B + renamer.addConstraint(lastDimA, secondLastDimB, DimensionRenamer.Constraint.equal(false), this); - assertTwoDimensions(aDimensions, inputs.get(0), "first argument"); - assertTwoDimensions(bDimensions, inputs.get(1), "second argument"); + // For efficiency, the dimensions to join over should be innermost - soft constraint + if (typeA.rank() >= 2) { + renamer.addConstraint(secondLastDimA, lastDimA, DimensionRenamer.Constraint.lessThan(true), this); + } + if (typeB.rank() >= 2) { + renamer.addConstraint(secondLastDimB, lastDimB, DimensionRenamer.Constraint.greaterThan(true), this); + } - String aDim0 = aDimensions.get(0).name(); - String aDim1 = aDimensions.get(1).name(); - String bDim0 = bDimensions.get(0).name(); - String bDim1 = bDimensions.get(1).name(); + // The second-to-last dimension of a should have a different name than the last dimension of b + if (typeA.rank() >= 2 && typeB.rank() >= 2) { + renamer.addConstraint(secondLastDimA, lastDimB, DimensionRenamer.Constraint.lessThan(false), this); + } - // The second dimension of a should have the same name as the first dimension of b - renamer.addConstraint(aDim1, bDim0, DimensionRenamer.Constraint.equal(false), this); + // a1 < a2 < a3 < a4 + OrderedTensorType largestRankType = typeA.rank() >= typeB.rank() ? typeA : typeB; + for (int i = 0; i < largestRankType.rank() - 2; ++i) { + String iDim = largestRankType.dimensionNames().get(i); + for (int j = i+1; j < largestRankType.rank() - 2; ++j) { + String jDim = largestRankType.dimensionNames().get(j); + renamer.addConstraint(iDim, jDim, DimensionRenamer.Constraint.lessThan(true), this); + } + } + + // TODO: handle non similar sizes + + // a1 == b1 etc + if (typeA.rank() == typeB.rank()) { + for (int i = 0; i < typeA.rank() - 2; ++i) { + renamer.addConstraint(typeA.dimensionNames().get(i), typeB.dimensionNames().get(i), DimensionRenamer.Constraint.equal(false), this); + } + } - // The first dimension of a should have a different name than the second dimension of b - renamer.addConstraint(aDim0, bDim1, DimensionRenamer.Constraint.lessThan(false), this); - // For efficiency, the dimensions to join over should be innermost - soft constraint - renamer.addConstraint(aDim0, aDim1, DimensionRenamer.Constraint.lessThan(true), this); - renamer.addConstraint(bDim0, bDim1, DimensionRenamer.Constraint.greaterThan(true), this); - } - private void assertTwoDimensions(List<TensorType.Dimension> dimensions, IntermediateOperation supplier, String inputDescription) { - if (dimensions.size() >= 2) return; - throw new IllegalArgumentException("Expected 2 dimensions in the " + inputDescription + " to " + this + - " but got just " + dimensions + " from\n" + - ExpressionFormatter.inTwoColumnMode(70, 50).format(supplier.toFullString())); + + // So, what about the other dimensions? +// if (aDimensions.size() > 2) { +// for (int i = 1; i < aDimensions.size(); ++i) { +// renamer.addConstraint(aDimensions.get(0).name(), aDimensions.get(i).name(), DimensionRenamer.Constraint.notEqual(false), this); +// } +// for (int i = 0; i < bDimensions.size(); ++i) { +// renamer.addConstraint(aDimensions.get(0).name(), bDimensions.get(i).name(), DimensionRenamer.Constraint.notEqual(false), this); +// } +// } + } +// private void assertTwoDimensions(List<TensorType.Dimension> dimensions, IntermediateOperation supplier, String inputDescription) { +// if (dimensions.size() >= 2) return; +// throw new IllegalArgumentException("Expected 2 dimensions in the " + inputDescription + " to " + this + +// " but got just " + dimensions + " from\n" + +// ExpressionFormatter.inTwoColumnMode(70, 50).format(supplier.toFullString())); +// } + @Override public MatMul withInputs(List<IntermediateOperation> inputs) { return new MatMul(modelName(), name(), inputs); diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Rename.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Rename.java index e040ae62149..07ac457cca8 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Rename.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Rename.java @@ -54,7 +54,7 @@ public class Rename extends IntermediateOperation { } public void renameDimensions(DimensionRenamer renamer) { - type = type.rename(renamer); + super.renameDimensions(renamer); from = renamer.dimensionNameOf(from).orElse(from); to = renamer.dimensionNameOf(to).orElse(to); } diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Reshape.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Reshape.java index c88fc18e6c6..f96dd420d30 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Reshape.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Reshape.java @@ -2,8 +2,10 @@ package ai.vespa.rankingexpression.importer.operations; import ai.vespa.rankingexpression.importer.OrderedTensorType; +import com.yahoo.searchlib.rankingexpression.Reference; import com.yahoo.searchlib.rankingexpression.evaluation.DoubleValue; import ai.vespa.rankingexpression.importer.DimensionRenamer; +import com.yahoo.searchlib.rankingexpression.evaluation.StringValue; import com.yahoo.searchlib.rankingexpression.evaluation.Value; import com.yahoo.searchlib.rankingexpression.rule.ArithmeticNode; import com.yahoo.searchlib.rankingexpression.rule.ArithmeticOperator; @@ -11,8 +13,11 @@ import com.yahoo.searchlib.rankingexpression.rule.ComparisonNode; import com.yahoo.searchlib.rankingexpression.rule.ConstantNode; import com.yahoo.searchlib.rankingexpression.rule.EmbracedNode; import com.yahoo.searchlib.rankingexpression.rule.ExpressionNode; +import com.yahoo.searchlib.rankingexpression.rule.Function; +import com.yahoo.searchlib.rankingexpression.rule.FunctionNode; import com.yahoo.searchlib.rankingexpression.rule.GeneratorLambdaFunctionNode; import com.yahoo.searchlib.rankingexpression.rule.ReferenceNode; +import com.yahoo.searchlib.rankingexpression.rule.TensorFunctionNode; import com.yahoo.searchlib.rankingexpression.rule.TruthOperator; import com.yahoo.tensor.Tensor; import com.yahoo.tensor.TensorType; @@ -27,6 +32,8 @@ import java.util.List; import java.util.Optional; import java.util.stream.Collectors; +import static com.yahoo.searchlib.rankingexpression.rule.TensorFunctionNode.wrapScalar; + public class Reshape extends IntermediateOperation { private final AttributeMap attributeMap; @@ -38,6 +45,10 @@ public class Reshape extends IntermediateOperation { @Override protected OrderedTensorType lazyGetType() { + + // required as we use tensor create + inputs.get(0).exportAsRankingFunction = true; + if (inputs.size() == 2) { return typeWithShapeAsInput(); } else if (inputs.size() == 1) { @@ -126,10 +137,54 @@ public class Reshape extends IntermediateOperation { return new Reshape(modelName(), name(), inputs, attributeMap); } - public static TensorFunction reshape(TensorFunction inputFunction, OrderedTensorType inputType, OrderedTensorType outputType) { + public TensorFunction reshape(TensorFunction inputFunction, OrderedTensorType inputType, OrderedTensorType outputType) { if ( ! OrderedTensorType.tensorSize(inputType.type()).equals(OrderedTensorType.tensorSize(outputType.type()))) throw new IllegalArgumentException("New and old shape of tensor must have the same size when reshaping"); + IntermediateOperation input = inputs.get(0); + String inputFunctionName = input.rankingExpressionFunctionName(); + + List<com.yahoo.tensor.functions.Slice.DimensionValue<Reference>> dimensionValues = new ArrayList<>(); + + // ala (d0 * 2 + d1) + ExpressionNode unrolled = new EmbracedNode(unrollTensorExpression(outputType)); + + long innerSize = 1; + for (int dim = 0; dim < inputType.rank(); ++dim) { + innerSize *= inputType.dimensions().get(dim).size().get(); + } + + for (int dim = 0; dim < inputType.rank(); ++dim) { + String inputDimensionName = inputType.dimensions().get(dim).name(); + long inputDimensionSize = inputType.dimensions().get(dim).size().get(); + long previousInnerSize = innerSize; + innerSize /= inputDimensionSize; + + ExpressionNode inputDimensionExpression; + if (inputDimensionSize == 1) { + inputDimensionExpression = new EmbracedNode(new ConstantNode(DoubleValue.zero)); + } else if (dim == (inputType.rank() - 1)) { + ExpressionNode size = new ConstantNode(new DoubleValue(inputDimensionSize)); + ExpressionNode div = new ArithmeticNode(unrolled, ArithmeticOperator.MODULO, size); + inputDimensionExpression = new EmbracedNode(div); + } else { + ExpressionNode size = new ConstantNode(new DoubleValue(innerSize)); + ExpressionNode previousSize = new ConstantNode(new DoubleValue(previousInnerSize)); + ExpressionNode mod = new ArithmeticNode(unrolled, ArithmeticOperator.MODULO, previousSize); + ExpressionNode div = new ArithmeticNode(new EmbracedNode(mod), ArithmeticOperator.DIVIDE, size); + inputDimensionExpression = new EmbracedNode(new FunctionNode(Function.floor, div)); + } + dimensionValues.add(new com.yahoo.tensor.functions.Slice.DimensionValue<>(Optional.of(inputDimensionName), wrapScalar(inputDimensionExpression))); + } + + TensorFunction<Reference> inputIndices = new TensorFunctionNode.ExpressionTensorFunction(new ReferenceNode(inputFunctionName)); + com.yahoo.tensor.functions.Slice<Reference> sliceIndices = new com.yahoo.tensor.functions.Slice<>(inputIndices, dimensionValues); + ExpressionNode sliceExpression = new TensorFunctionNode(sliceIndices); + + TensorFunction generate = Generate.bound(outputType.type(), wrapScalar(sliceExpression)); + return generate; + + /* // Conceptually, reshaping consists on unrolling a tensor to an array using the dimension order, // then use the dimension order of the new shape to roll back into a tensor. // Here we create a transformation tensor that is multiplied with the from tensor to map into @@ -168,11 +223,14 @@ public class Reshape extends IntermediateOperation { result = new Rename(result, to, from); } return result; + */ } + /* private static boolean dimensionNamesOverlap(OrderedTensorType a, OrderedTensorType b) { return a.dimensionNames().stream().anyMatch(d -> b.type().indexOfDimension(d).isPresent()); } + */ private static ExpressionNode unrollTensorExpression(OrderedTensorType type) { if (type.rank() == 0) diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Slice.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Slice.java index e5463291ef8..8dd1e3ff33d 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Slice.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Slice.java @@ -182,7 +182,6 @@ public class Slice extends IntermediateOperation { @Override public void addDimensionNameConstraints(DimensionRenamer renamer) { - // Todo: what to do? for (int i = 0; i < type.dimensions().size(); i++) { renamer.addDimension(type.dimensions().get(i).name()); for (int j = i + 1; j < type.dimensions().size(); j++) { diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Softmax.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Softmax.java index 83086926316..e2b83246bfc 100644 --- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Softmax.java +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Softmax.java @@ -5,6 +5,7 @@ import ai.vespa.rankingexpression.importer.OrderedTensorType; import com.yahoo.tensor.functions.Join; import com.yahoo.tensor.functions.Map; import com.yahoo.tensor.functions.Reduce; +import com.yahoo.tensor.functions.ScalarFunction; import com.yahoo.tensor.functions.ScalarFunctions; import com.yahoo.tensor.functions.TensorFunction; @@ -28,6 +29,10 @@ public class Softmax extends IntermediateOperation { @Override protected OrderedTensorType lazyGetType() { if ( ! allInputTypesPresent(1)) return null; + + // input is referenced twice due to avoidance of overflow. so make this it's own function. + inputs.get(0).exportAsRankingFunction = true; + return inputs.get(0).type().get(); } @@ -50,7 +55,9 @@ public class Softmax extends IntermediateOperation { } TensorFunction input = inputs.get(0).function().get(); - TensorFunction exp = new Map(input, ScalarFunctions.exp()); + TensorFunction max = new Reduce(input, Reduce.Aggregator.max, reduceDimensions); + TensorFunction cap = new Join(input, max, ScalarFunctions.subtract()); // to avoid overflow + TensorFunction exp = new Map(cap, ScalarFunctions.exp()); TensorFunction sum = new Reduce(exp, Reduce.Aggregator.sum, reduceDimensions); TensorFunction div = new Join(exp, sum, ScalarFunctions.divide()); diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Split.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Split.java new file mode 100644 index 00000000000..02d780c52cd --- /dev/null +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Split.java @@ -0,0 +1,119 @@ +// Copyright 2020 Oath Inc. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. +package ai.vespa.rankingexpression.importer.operations; + +import ai.vespa.rankingexpression.importer.OrderedTensorType; +import com.yahoo.searchlib.rankingexpression.Reference; +import com.yahoo.searchlib.rankingexpression.evaluation.DoubleValue; +import com.yahoo.searchlib.rankingexpression.evaluation.Value; +import com.yahoo.searchlib.rankingexpression.rule.ArithmeticNode; +import com.yahoo.searchlib.rankingexpression.rule.ArithmeticOperator; +import com.yahoo.searchlib.rankingexpression.rule.ConstantNode; +import com.yahoo.searchlib.rankingexpression.rule.EmbracedNode; +import com.yahoo.searchlib.rankingexpression.rule.ExpressionNode; +import com.yahoo.searchlib.rankingexpression.rule.ReferenceNode; +import com.yahoo.searchlib.rankingexpression.rule.TensorFunctionNode; +import com.yahoo.tensor.TensorType; +import com.yahoo.tensor.functions.Generate; +import com.yahoo.tensor.functions.TensorFunction; + +import java.util.ArrayList; +import java.util.List; +import java.util.Optional; + +import static com.yahoo.searchlib.rankingexpression.rule.TensorFunctionNode.wrapScalar; + +public class Split extends IntermediateOperation { + + private final AttributeMap attributes; + private final int output; + + private final int axis; + private int start; + private int end; + + public Split(String modelName, String nodeName, List<IntermediateOperation> inputs, AttributeMap attributes, int output) { + super(modelName, nodeName, inputs); + this.attributes = attributes; + this.output = output; + axis = (int) attributes.get("axis").orElse(DoubleValue.zero).asDouble(); + } + + @Override + protected OrderedTensorType lazyGetType() { + if (!allInputTypesPresent(1)) + return null; + OrderedTensorType inputType = inputs.get(0).type().get(); + + // required as we use tensor create + inputs.get(0).exportAsRankingFunction = true; + + int axisSize = inputType.dimensions().get(axis).size().get().intValue(); + start = 0; + end = axisSize; + + if (attributes.getList("split").isPresent()) { + List<Value> splitList = attributes.getList("split").get(); + if (output > splitList.size()) { + throw new IllegalArgumentException("Split in " + name + ": output out of range of split list"); + } + for (int i = 0; i < output; ++i) { + start += (int) splitList.get(i).asDouble(); + } + if (output < splitList.size()) { + end = start + (int) splitList.get(output).asDouble(); + } + } else { + start = axisSize / 2 * output; + end = start + axisSize / 2; + } + + if (start >= axisSize || start < 0) { + throw new IllegalArgumentException("Split in " + name + ": split start index out of range (" + start + ")"); + } + if (end > axisSize || end < 0) { + throw new IllegalArgumentException("Split in " + name + ": split end index out of range (" + end + ")"); + } + + OrderedTensorType.Builder typeBuilder = new OrderedTensorType.Builder(resultValueType()); + for (int i = 0; i < inputType.rank(); ++i) { + TensorType.Dimension inputDimension = inputType.dimensions().get(i); + long dimSize = i == axis ? end - start : inputDimension.size().get(); + typeBuilder.add(TensorType.Dimension.indexed(inputDimension.name(), dimSize)); + } + return typeBuilder.build(); + } + + @Override + protected TensorFunction lazyGetFunction() { + if (!allInputFunctionsPresent(1)) return null; + + IntermediateOperation input = inputs.get(0); + OrderedTensorType inputType = input.type().get(); + String inputFunctionName = input.rankingExpressionFunctionName(); + + List<com.yahoo.tensor.functions.Slice.DimensionValue<Reference>> dimensionValues = new ArrayList<>(); + + for (int i = 0; i < inputType.rank(); ++i) { + String inputDimensionName = inputType.dimensions().get(i).name(); + ExpressionNode reference = new ReferenceNode(inputDimensionName); + ExpressionNode offset = new ArithmeticNode(reference, ArithmeticOperator.PLUS, new ConstantNode(new DoubleValue(i == axis ? start : 0))); + dimensionValues.add(new com.yahoo.tensor.functions.Slice.DimensionValue<>(Optional.of(inputDimensionName), wrapScalar(new EmbracedNode(offset)))); + } + + TensorFunction<Reference> inputIndices = new TensorFunctionNode.ExpressionTensorFunction(new ReferenceNode(inputFunctionName)); + com.yahoo.tensor.functions.Slice<Reference> sliceIndices = new com.yahoo.tensor.functions.Slice<>(inputIndices, dimensionValues); + ExpressionNode sliceExpression = new TensorFunctionNode(sliceIndices); + + TensorFunction generate = Generate.bound(type.type(), wrapScalar(sliceExpression)); + return generate; + } + + @Override + public Split withInputs(List<IntermediateOperation> inputs) { + return new Split(modelName(), name(), inputs, attributes, output); + } + + @Override + public String operationName() { return "Split"; } + +} diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Tile.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Tile.java new file mode 100644 index 00000000000..8d3468f3d04 --- /dev/null +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Tile.java @@ -0,0 +1,100 @@ +// Copyright 2020 Oath Inc. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. +package ai.vespa.rankingexpression.importer.operations; + +import ai.vespa.rankingexpression.importer.OrderedTensorType; +import com.yahoo.searchlib.rankingexpression.Reference; +import com.yahoo.searchlib.rankingexpression.evaluation.DoubleValue; +import com.yahoo.searchlib.rankingexpression.rule.ArithmeticNode; +import com.yahoo.searchlib.rankingexpression.rule.ArithmeticOperator; +import com.yahoo.searchlib.rankingexpression.rule.ConstantNode; +import com.yahoo.searchlib.rankingexpression.rule.EmbracedNode; +import com.yahoo.searchlib.rankingexpression.rule.ExpressionNode; +import com.yahoo.searchlib.rankingexpression.rule.ReferenceNode; +import com.yahoo.searchlib.rankingexpression.rule.TensorFunctionNode; +import com.yahoo.tensor.Tensor; +import com.yahoo.tensor.TensorType; +import com.yahoo.tensor.functions.Generate; +import com.yahoo.tensor.functions.TensorFunction; + +import java.util.ArrayList; +import java.util.List; +import java.util.Optional; + +import static com.yahoo.searchlib.rankingexpression.rule.TensorFunctionNode.wrapScalar; + +/** + * Onnx tile operation. + */ +public class Tile extends IntermediateOperation { + + public Tile(String modelName, String nodeName, List<IntermediateOperation> inputs) { + super(modelName, nodeName, inputs); + } + + @Override + protected OrderedTensorType lazyGetType() { + if (!allInputTypesPresent(2)) return null; + + // required as we use tensor create + inputs.get(0).exportAsRankingFunction = true; + + IntermediateOperation repeats = inputs.get(1); + if (repeats.getConstantValue().isEmpty()) + throw new IllegalArgumentException("Tile " + name + ": repeats input must be a constant."); + + Tensor shape = repeats.getConstantValue().get().asTensor(); + if (shape.type().rank() != 1) + throw new IllegalArgumentException("Tile " + name + ": repeats must be a 1-d tensor."); + + OrderedTensorType inputType = inputs.get(0).type().get(); + if (shape.type().dimensions().get(0).size().get() != inputType.rank()) + throw new IllegalArgumentException("Tile " + name + ": repeats must be the same size as input rank."); + + List<Integer> dimSizes = new ArrayList<>(inputType.rank()); + shape.valueIterator().forEachRemaining(v -> dimSizes.add(v.intValue())); + + OrderedTensorType.Builder typeBuilder = new OrderedTensorType.Builder(resultValueType()); + for (int i = 0; i < dimSizes.size(); ++i) { + TensorType.Dimension inputDimension = inputType.dimensions().get(i); + typeBuilder.add(TensorType.Dimension.indexed(inputDimension.name(), inputDimension.size().get() * dimSizes.get(i))); + } + return typeBuilder.build(); + } + + @Override + protected TensorFunction lazyGetFunction() { + if (!allInputFunctionsPresent(2)) return null; + + IntermediateOperation input = inputs.get(0); + OrderedTensorType inputType = input.type().get(); + String inputFunctionName = input.rankingExpressionFunctionName(); + + List<com.yahoo.tensor.functions.Slice.DimensionValue<Reference>> dimensionValues = new ArrayList<>(); + + for (int axis = 0; axis < inputType.rank(); ++axis) { + String inputDimensionName = inputType.dimensions().get(axis).name(); + long inputDimensionSize = inputType.dimensions().get(axis).size().get(); + + ExpressionNode size = new ConstantNode(new DoubleValue(inputDimensionSize)); + ExpressionNode reference = new ReferenceNode(inputDimensionName); + ExpressionNode mod = new ArithmeticNode(reference, ArithmeticOperator.MODULO, size); + dimensionValues.add(new com.yahoo.tensor.functions.Slice.DimensionValue<>(Optional.of(inputDimensionName), wrapScalar(new EmbracedNode(mod)))); + } + + TensorFunction<Reference> inputIndices = new TensorFunctionNode.ExpressionTensorFunction(new ReferenceNode(inputFunctionName)); + com.yahoo.tensor.functions.Slice<Reference> sliceIndices = new com.yahoo.tensor.functions.Slice<>(inputIndices, dimensionValues); + ExpressionNode sliceExpression = new TensorFunctionNode(sliceIndices); + + TensorFunction generate = Generate.bound(type.type(), wrapScalar(sliceExpression)); + return generate; + } + + @Override + public Tile withInputs(List<IntermediateOperation> inputs) { + return new Tile(modelName(), name(), inputs); + } + + @Override + public String operationName() { return "Tile"; } + +} diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Transpose.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Transpose.java new file mode 100644 index 00000000000..178759fbf2a --- /dev/null +++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/operations/Transpose.java @@ -0,0 +1,54 @@ +// Copyright 2018 Yahoo Holdings. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. +package ai.vespa.rankingexpression.importer.operations; + +import ai.vespa.rankingexpression.importer.OrderedTensorType; +import com.yahoo.tensor.Tensor; +import com.yahoo.tensor.TensorType; +import com.yahoo.tensor.functions.TensorFunction; + +import java.util.List; + +public class Transpose extends IntermediateOperation { + + private final AttributeMap attributes; + + public Transpose(String modelName, String nodeName, List<IntermediateOperation> inputs, AttributeMap attributes) { + super(modelName, nodeName, inputs); + this.attributes = attributes; + } + + @Override + protected OrderedTensorType lazyGetType() { + if (!allInputTypesPresent(1)) return null; + + OrderedTensorType inputType = inputs.get(0).type().get(); + + OrderedTensorType.Builder typeBuilder = new OrderedTensorType.Builder(resultValueType()); + for (int i = 0; i < inputType.rank(); ++i) { + int inputIndex = inputType.rank() - 1 - i; + if (attributes.getList("perm").isPresent()) { + inputIndex = (int) attributes.getList("perm").get().get(i).asDouble(); + } + TensorType.Dimension inputDimension = inputType.dimensions().get(inputIndex); + typeBuilder.add(TensorType.Dimension.indexed(inputDimension.name(), inputDimension.size().get())); + } + OrderedTensorType result = typeBuilder.build(); + return typeBuilder.build(); + } + + @Override + protected TensorFunction lazyGetFunction() { + if (!allInputFunctionsPresent(1)) + return null; + return inputs.get(0).function().orElse(null); + } + + @Override + public Transpose withInputs(List<IntermediateOperation> inputs) { + return new Transpose(modelName(), name(), inputs, attributes); + } + + @Override + public String operationName() { return "Transpose"; } + +} diff --git a/model-integration/src/test/java/ai/vespa/rankingexpression/importer/onnx/BertImportTestCase.java b/model-integration/src/test/java/ai/vespa/rankingexpression/importer/onnx/BertImportTestCase.java new file mode 100644 index 00000000000..f4ed2f1b64d --- /dev/null +++ b/model-integration/src/test/java/ai/vespa/rankingexpression/importer/onnx/BertImportTestCase.java @@ -0,0 +1,281 @@ +// Copyright 2019 Oath Inc. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. +package ai.vespa.rankingexpression.importer.onnx; + +import ai.vespa.rankingexpression.importer.ImportedModel; +import com.yahoo.io.IOUtils; +import com.yahoo.searchlib.rankingexpression.RankingExpression; +import com.yahoo.searchlib.rankingexpression.Reference; +import com.yahoo.searchlib.rankingexpression.evaluation.Context; +import com.yahoo.searchlib.rankingexpression.evaluation.ContextIndex; +import com.yahoo.searchlib.rankingexpression.evaluation.DoubleValue; +import com.yahoo.searchlib.rankingexpression.evaluation.MapContext; +import com.yahoo.searchlib.rankingexpression.evaluation.TensorValue; +import com.yahoo.searchlib.rankingexpression.evaluation.Value; +import com.yahoo.searchlib.rankingexpression.parser.ParseException; +import com.yahoo.searchlib.rankingexpression.rule.CompositeNode; +import com.yahoo.searchlib.rankingexpression.rule.ExpressionNode; +import com.yahoo.searchlib.rankingexpression.rule.ReferenceNode; +import com.yahoo.searchlib.rankingexpression.rule.TensorFunctionNode; +import com.yahoo.tensor.Tensor; +import com.yahoo.tensor.TensorType; +import com.yahoo.tensor.functions.Generate; +import com.yahoo.tensor.functions.ScalarFunction; +import com.yahoo.tensor.functions.Slice; +import org.junit.Ignore; +import org.junit.Test; +import org.tensorflow.op.core.Rank; + +import java.io.BufferedReader; +import java.io.IOException; +import java.sql.Ref; +import java.util.HashSet; +import java.util.List; +import java.util.Map; +import java.util.Set; + +import static org.junit.Assert.assertEquals; + +/** + * @author lesters + */ +public class BertImportTestCase extends TestableModel { + + @Test + public void test() throws Exception { + String filename = "/Users/lesters/github/onnx-models/text/machine_comprehension/bert-squad/java.txt"; + List<String> lines = IOUtils.getLines(filename); + Tensor tensor = Tensor.from(lines.get(0)); + TestableModelContext context = new TestableModelContext(); + context.put("test", new TensorValue(tensor)); + + // Tensor: tensor(d1[1],d2[256],d4[12],d5[64]) + + String expr = "tensor(d0[256],d1[768])" + + "((test{" + + "d1:(floor(0.0)), " + + "d2:(floor(((768.0 * d0 + d1) % 196608) / 768.0)), " + + "d4:(floor(((768.0 * d0 + d1) % 768.0) / 64.0)), " + + "d5:(floor((768.0 * d0 + d1) % 64.0))" + + "}))"; + Tensor result = new RankingExpression(expr).evaluate(context).asTensor(); + + assertEquals(result.sum(), -6074.247); + } + + @Ignore + @Test + public void testBertImport() { + ImportedModel model = new OnnxImporter().importModel("test", "/Users/lesters/github/onnx-models/text/machine_comprehension/bert-squad/bertsquad8_modified.onnx"); +// ImportedModel model = new OnnxImporter().importModel("test", "src/test/models/onnx/bert/bertsquad8_modified.onnx"); +// ImportedModel model = new OnnxImporter().importModel("test", "src/test/models/onnx/bert/bertsquad10.onnx"); +// assertEquals(0, model.signature("default").skippedOutputs().size()); +// Tensor onnxResult = evaluateVespa(model, "output", model.inputs()); +// assertEquals(Tensor.from("tensor(d0[1],d1[2]):[[0.28258783057229725, -0.0685615853647904]]"), onnxResult); + + String filename = "/Users/lesters/github/onnx-models/text/machine_comprehension/bert-squad/context.vespa"; + + // bert/encoder/layer_0/attention/self/mul_2 + assert null != model.largeConstants().get("test_bert_encoder_layer_0_attention_self_Reshape_3__294"); + + TestableModelContext context; + if (true) { + // inputs + Tensor unique_ids_raw_output__9 = Tensor.from("tensor(d0[1]):[1]"); + Tensor input_ids = Tensor.from("tensor(d0[1],d1[256]):[101,2073,2003,1996,5661,10549,2000,2175,1029,102,1999,2049,2220,2086,1010,1996,2047,4680,2415,3478,2000,3113,5270,1998,6599,10908,1012,1031,2260,1033,2011,2526,1010,2116,13773,3028,5661,2020,10549,1996,2172,3469,9587,9363,2638,2415,1999,2624,3799,2058,1996,2624,4560,4680,2415,2349,2000,1996,3732,1005,1055,3132,2686,1012,1037,10428,5468,2000,5446,2019,4935,3081,1037,3309,4171,3478,2000,3362,1996,3223,2048,1011,12263,3484,2000,3413,1012,1999,2238,2384,1010,2136,2624,4560,2328,1996,2148,2534,1010,1037,1002,1020,1012,6255,2454,1010,2630,1998,2317,9311,1010,5815,3770,1010,2199,2675,2519,1006,1021,1010,4278,25525,1007,1997,8327,2686,102,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]"); + Tensor input_mask = Tensor.from("tensor(d0[1],d1[256]):[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]"); + Tensor segment_ids = Tensor.from("tensor(d0[1],d1[256]):[0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]"); + + context = contextFrom(model); + context.put("unique_ids_raw_output___9", new TensorValue(unique_ids_raw_output__9)); + context.put("input_ids", new TensorValue(input_ids)); + context.put("input_mask", new TensorValue(input_mask)); + context.put("segment_ids", new TensorValue(segment_ids)); + + context.write(filename); + } else { + context = TestableModelContext.read(filename); + } + + // expected outputs from onnxruntime + Tensor unique_ids = Tensor.from("tensor(d0[1]):[1]"); + Tensor unstack_0 = Tensor.from("tensor(d0[1],d1[256]):[-7.169589,-8.165145,-8.795558,-8.276284,-8.408593,-8.313643,-8.421538,-8.771402,-8.71111,-8.014886,-6.4415646,-7.5764513,-7.7209125,-8.689668,-8.05441,-4.357495,-4.082243,-2.4557219,-6.421309,-7.8627315,-8.612887,-8.122109,-8.072487,-8.678347,-8.467162,-8.818881,-8.143695,-6.412044,-7.765201,-8.125683,-2.5739796,-4.2929254,-7.8812947,-2.4893682,-1.5166948,-6.2354407,-4.039099,-5.6837378,-0.41342238,3.0958412,1.5454307,-0.89450985,6.0985346,-4.108738,-4.67186,-2.6797965,-0.65007347,4.2300944,-2.9132476,-4.853151,1.1584995,4.041984,-3.5257776,-2.3050616,-4.363427,-7.1510825,-8.426602,-6.6682553,-7.027374,-8.076435,-8.3017435,-6.9958987,-7.243815,-7.1347113,-7.5506253,-7.771371,-8.606251,-7.472072,-7.902196,-7.563202,-7.330995,-7.5767503,-7.8097973,-6.645113,-8.927777,-8.438513,-8.708496,-8.474434,-8.231956,-8.635139,-7.8764973,-8.80273,-9.103729,-9.07057,-8.610826,-9.084642,-8.795743,-7.2711506,-7.733648,-8.708181,-8.020964,-7.1652384,-7.9469404,-9.461184,-8.624146,-7.2252526,-6.4015207,-9.220176,-9.195709,-8.228707,-7.9646325,-8.685807,-8.980191,-8.858017,-9.290145,-8.921865,-7.656322,-8.872562,-8.898288,-8.683226,-9.219653,-8.371141,-7.130355,-8.930712,-9.05438,-8.771264,-9.621703,-8.550959,-7.327657,-9.138217,-9.377564,-9.111144,-9.653343,-8.726485,-8.215803,-9.300696,-8.044907,-8.641199,-8.641449,-8.640882,-8.641207,-8.648375,-8.645785,-8.639973,-8.650788,-8.660226,-8.6503525,-8.6601925,-8.647732,-8.652576,-8.665123,-8.653585,-8.653888,-8.661093,-8.661934,-8.6514845,-8.662573,-8.671499,-8.661195,-8.667901,-8.666959,-8.659721,-8.673244,-8.678537,-8.66441,-8.651034,-8.660175,-8.659063,-8.657169,-8.6603565,-8.6569,-8.649067,-8.651927,-8.6421995,-8.649052,-8.6478615,-8.6426935,-8.646153,-8.646865,-8.636821,-8.643324,-8.645994,-8.639597,-8.647679,-8.655649,-8.6609745,-8.654906,-8.6613455,-8.656511,-8.663024,-8.675192,-8.663131,-8.665018,-8.652522,-8.661668,-8.66894,-8.670112,-8.67217,-8.657303,-8.651893,-8.652592,-8.650168,-8.640702,-8.636455,-8.647628,-8.638621,-8.648,-8.656844,-8.649821,-8.657603,-8.648884,-8.661986,-8.663507,-8.652322,-8.662775,-8.664504,-8.662872,-8.668943,-8.6559105,-8.655738,-8.671845,-8.6666,-8.659552,-8.679308,-8.659756,-8.664594,-8.6688175,-8.666396,-8.673796,-8.65924,-8.664916,-8.6703005,-8.6611395,-8.660061,-8.660967,-8.672797,-8.66394,-8.657039,-8.671023,-8.663469,-8.659371,-8.6713705,-8.659359,-8.649764,-8.6620035,-8.656843,-8.654225,-8.661666,-8.647326,-8.652874,-8.650523,-8.644273,-8.649993,-8.65307,-8.645219,-8.6537075,-8.655814,-8.654312,-8.658724,-8.666763,-8.654713,-8.662302,-8.672376,-8.661079,-8.659652,-8.661736]"); + Tensor unstack_1 = Tensor.from("tensor(d0[1],d1[256]):[-5.1743593,-8.167716,-8.096918,-8.610186,-8.627197,-8.518608,-8.413071,-8.04796,-8.405228,-5.775467,-8.891069,-8.499419,-8.482899,-7.4575906,-5.1060586,-9.029796,-7.9796743,-7.411322,-0.62632525,-8.209348,-8.202109,-8.436105,-8.226212,-8.245562,-7.7150273,-5.4672513,-6.134469,-8.531252,-7.390566,-6.717802,-7.9110403,-5.084878,-5.02966,-7.6901536,-7.6643076,-0.42670453,-4.2289968,-6.957412,-5.192218,-6.1616683,-6.4489427,-3.5914042,-3.7853065,-6.857571,-2.3781726,6.1620126,-3.007885,-4.688912,6.258016,-5.2202945,-6.7945094,-5.1450105,0.7468612,-4.919924,5.489712,-7.307814,-7.952,-9.152897,-6.4863043,-8.328119,-7.9448185,-8.395245,-3.6581624,-0.8252581,-8.731679,-8.624653,-7.61354,-8.755644,-8.341698,-8.758186,-5.954141,-8.560192,-8.833243,-7.6137505,-5.96118,-8.43961,-8.188338,-8.373185,-8.683964,-8.246368,-8.824446,-8.05728,-7.623751,-7.56998,-8.277908,-6.8986,-6.8709283,-9.279125,-8.84588,-7.6791453,-5.2976,-9.191589,-8.797903,-6.440836,-8.179676,-9.236156,-8.972708,-5.8724217,-6.928253,-8.685118,-8.84946,-8.293621,-7.8572874,-8.053903,-7.398021,-7.549705,-9.004784,-8.060446,-7.950672,-7.2188964,-6.497633,-8.454956,-9.045556,-7.8463507,-7.771165,-8.067679,-5.9176393,-8.09684,-8.5619955,-7.5696144,-7.100621,-7.0136676,-6.464568,-8.108538,-8.516457,-5.488856,-5.853514,-8.457255,-8.457188,-8.454844,-8.448273,-8.447864,-8.447953,-8.445874,-8.444903,-8.442595,-8.448225,-8.443758,-8.451776,-8.447646,-8.440473,-8.447313,-8.44705,-8.443977,-8.442994,-8.449743,-8.441086,-8.433916,-8.43898,-8.435363,-8.434594,-8.431777,-8.433416,-8.433545,-8.442987,-8.453411,-8.450068,-8.4503565,-8.451651,-8.450909,-8.454222,-8.456041,-8.452284,-8.449699,-8.454986,-8.455096,-8.459543,-8.458114,-8.458371,-8.4632635,-8.458183,-8.457299,-8.458008,-8.452067,-8.444335,-8.442348,-8.445211,-8.441855,-8.443939,-8.441303,-8.436119,-8.442878,-8.439337,-8.446676,-8.441184,-8.438475,-8.440033,-8.4386015,-8.447922,-8.455316,-8.452563,-8.454967,-8.459164,-8.460839,-8.453004,-8.451543,-8.446279,-8.441412,-8.448481,-8.446184,-8.448539,-8.445241,-8.444487,-8.450539,-8.446448,-8.446319,-8.447268,-8.440758,-8.448286,-8.447366,-8.437631,-8.441085,-8.444475,-8.431786,-8.441355,-8.436929,-8.432141,-8.436456,-8.435032,-8.445299,-8.442143,-8.438964,-8.445743,-8.445099,-8.444958,-8.438029,-8.439503,-8.446831,-8.43919,-8.442334,-8.446472,-8.442076,-8.449043,-8.451941,-8.449556,-8.454564,-8.455859,-8.452123,-8.461076,-8.45802,-8.456931,-8.458485,-8.45496,-8.4508295,-8.453123,-8.451649,-8.451098,-8.450148,-8.446929,-8.44253,-8.44839,-8.444667,-8.437894,-8.444409,-8.444666,-8.441956]"); + + +// model.functions().forEach((k, v) -> { +// evaluateFunction(context, model, k, ""); +// }); + +// RankingExpression e = model.expressions().get("unique_ids_graph_outputs_Identity__10"); +// evaluateFunctionDependencies(context, model, e.getRoot(), ""); +// Tensor result = e.evaluate(context).asTensor(); +// assertEquals(result, unique_ids); + + RankingExpression e = model.expressions().get("bert/encoder/layer_0/output/LayerNorm/batchnorm/add_1"); + + evaluateFunctionDependencies(context, model, e.getRoot(), ""); + context.write(filename); + Tensor result = e.evaluate(context).asTensor(); + double sum = result.sum().asDouble(); + System.out.println(sum); + + Tensor matmul1 = model.expressions().get("bert/encoder/layer_0/attention/self/MatMul_1").evaluate(context).asTensor(); + + Tensor transpose = model.expressions().get("bert/encoder/layer_0/attention/self/transpose_3").evaluate(context).asTensor(); + String cast = model.largeConstants().get("test_bert_encoder_layer_0_attention_self_Reshape_3__294"); + Tensor reshape = model.expressions().get("bert/encoder/layer_0/attention/self/Reshape_3").evaluate(context).asTensor(); + Tensor matmul = model.expressions().get("bert/encoder/layer_0/attention/output/dense/MatMul").evaluate(context).asTensor(); + Tensor add = model.expressions().get("bert/encoder/layer_0/attention/output/dense/BiasAdd").evaluate(context).asTensor(); + + Tensor add1 = model.expressions().get("bert/encoder/layer_0/attention/output/add").evaluate(context).asTensor(); + Tensor add2 = model.expressions().get("bert/encoder/layer_0/attention/output/LayerNorm/batchnorm/add_1").evaluate(context).asTensor(); + Tensor add3 = model.expressions().get("bert/encoder/layer_0/output/add").evaluate(context).asTensor(); + Tensor add4 = model.expressions().get("bert/encoder/layer_0/output/LayerNorm/batchnorm/add_1").evaluate(context).asTensor(); + + assertEquals(result, unique_ids); + +// Tensor result = model.expressions().get("unique_ids_graph_outputs_Identity__10").evaluate(context).asTensor(); +// assertEquals(result, unique_ids); + +// result = model.expressions().get("unstack_graph_outputs_Identity__7").evaluate(context).asTensor(); // or map from signature outputs +// assertEquals(result, unstack_0); + + // en feil her i outputs: har bare en: unstack, men vi mÃ¥ ha to: unstack:0 og unstack:1 + + } + + + private void evaluateFunction(Context context, ImportedModel model, String functionName, String in) { + if (!context.names().contains(functionName)) { + RankingExpression e = RankingExpression.from(model.functions().get(functionName)); + System.out.println(in + "Looking for dependencies of function " + functionName + ": " + e.toString()); + evaluateFunctionDependencies(context, model, e.getRoot(), in); + System.out.println(in + "Evaluating function " + functionName + ": " + e.toString()); + long start = System.currentTimeMillis(); + Tensor result = e.evaluate(context).asTensor(); + context.put(functionName, new TensorValue(result)); + long end = System.currentTimeMillis(); + System.out.println(in + "[" + (end - start) + "] completed " + functionName + " (" + result.type() + "), context is: " + context.names().size() + " " + contextSize(context)); + } else { + System.out.println(in + "Function " + functionName + " already evaluated..."); + } + } + + private long contextSize(Context context) { + long size = 0; + for (String name : context.names()) { + Tensor val = context.getTensor(name); + if (val != null) size += val.size(); + } + return size; + } + + private void evaluateFunctionDependencies(Context context, ImportedModel model, ExpressionNode node, String in) { + if (node instanceof ReferenceNode) { + String name = node.toString(); + ReferenceNode ref = (ReferenceNode) node; + if (ref.getName().equals("constant")) { + String constant = ref.getArguments().expressions().get(0).toString(); + if (!context.names().contains(constant)) { + String value = null; + if (model.smallConstants().containsKey(constant)) { + value = model.smallConstants().get(constant); + } + if (model.largeConstants().containsKey(constant)) { + value = model.largeConstants().get(constant); + } + if (value != null) { + System.out.println(in + "Adding constant: " + name); + long start = System.currentTimeMillis(); + Tensor val = Tensor.from(value); + context.put(name, new TensorValue(val)); + long end = System.currentTimeMillis(); + System.out.println(in + "Added constant: " + name + " (" + val.type() + ") in [" + (end - start) + "]"); + } + } + } + if (model.functions().containsKey(name)) { + evaluateFunction(context, model, name, in + " "); + } + } + else if (node instanceof CompositeNode) { + if (node instanceof TensorFunctionNode && ((TensorFunctionNode)node).function() instanceof Generate) { + Generate generate = (Generate) ((TensorFunctionNode)node).function(); + TensorFunctionNode.ExpressionScalarFunction func = (TensorFunctionNode.ExpressionScalarFunction) generate.getBoundGenerator(); + if (func != null) { + ExpressionNode bound = func.getExpression(); + if (bound.toString().contains("imported_ml_")) { + System.out.println(in + "Found expression inside generator: " + bound.toString()); + evaluateFunctionDependencies(context, model, bound, in); + } + } + } + else if (node instanceof TensorFunctionNode && ((TensorFunctionNode)node).function() instanceof Slice) { + Slice<Reference> slice = (Slice<Reference>) ((TensorFunctionNode)node).function(); + for (Slice.DimensionValue<Reference> value : slice.getSubspaceAddress()) { + TensorFunctionNode.ExpressionScalarFunction func = (TensorFunctionNode.ExpressionScalarFunction) value.index().orElse(null); + if (func != null) { + ExpressionNode bound = func.getExpression(); + if (bound.toString().contains("imported_ml_")) { + System.out.println(in + "Found expression inside slice: " + bound.toString()); + evaluateFunctionDependencies(context, model, bound, in); + } + } + } + } + for (ExpressionNode child : ((CompositeNode)node).children()) { + evaluateFunctionDependencies(context, model, child, in); + } + } + } + + static TestableModelContext contextFrom(ImportedModel result) { + TestableModelContext context = new TestableModelContext(); + if (result != null) { + result.largeConstants().forEach((name, tensor) -> context.put("constant(" + name + ")", new TensorValue(Tensor.from(tensor)))); + result.smallConstants().forEach((name, tensor) -> context.put("constant(" + name + ")", new TensorValue(Tensor.from(tensor)))); + } + return context; + } + + private static class TestableModelContext extends MapContext implements ContextIndex { + @Override + public int size() { + return bindings().size(); + } + @Override + public int getIndex(String name) { + throw new UnsupportedOperationException(this + " does not support index lookup by name"); + } + + public void write(String filename) { + try { + for (Map.Entry<String, Value> entry: bindings().entrySet()) { + String line = entry.getKey() + "\t" + entry.getValue().asTensor() + "\n"; + IOUtils.writeFile(filename, line, true); + } + } catch (IOException e) { + throw new RuntimeException(e); + } + } + + public static TestableModelContext read(String filename) { + System.out.println("Reading content from " + filename); + TestableModelContext context = new TestableModelContext(); + try (BufferedReader reader = IOUtils.createReader(filename)) { + String line; + while (null != (line = reader.readLine())) { + String[] strings = line.trim().split("\t"); + String name = strings[0]; + Tensor tensor = Tensor.from(strings[1]); + context.put(name, new TensorValue(tensor)); + } + } catch (IOException e) { + throw new RuntimeException(e); + } + System.out.println("Done reading context"); + return context; + } + } + +} diff --git a/model-integration/src/test/java/ai/vespa/rankingexpression/importer/onnx/OnnxOperationsTestCase.java b/model-integration/src/test/java/ai/vespa/rankingexpression/importer/onnx/OnnxOperationsTestCase.java index 94c5577357b..0c9acc9b372 100644 --- a/model-integration/src/test/java/ai/vespa/rankingexpression/importer/onnx/OnnxOperationsTestCase.java +++ b/model-integration/src/test/java/ai/vespa/rankingexpression/importer/onnx/OnnxOperationsTestCase.java @@ -107,6 +107,18 @@ public class OnnxOperationsTestCase { assertEval("less", x, y, evaluate("join(x, rename(y, d0, d2), f(a,b)(a < b))", x, y)); assertEval("equal", x, y, evaluate("join(x, rename(y, d0, d2), f(a,b)(a == b))", x, y)); assertEval("pow", x, y, evaluate("join(x, rename(y, d0, d2), f(a,b)(pow(a,b)))", x, y)); + + // broadcasting - opposite order + x = evaluate("random(d0[4]) + 1"); + y = evaluate("random(d0[2],d1[3],d2[4]) + 1"); + assertEval("add", x, y, evaluate("rename(x, d0, d2) + y", x, y)); + assertEval("sub", x, y, evaluate("rename(x, d0, d2) - y", x, y)); + assertEval("mul", x, y, evaluate("rename(x, d0, d2) * y", x, y)); + assertEval("div", x, y, evaluate("rename(x, d0, d2) / y", x, y)); + assertEval("greater", x, y, evaluate("join(rename(x, d0, d2), y, f(a,b)(a > b))", x, y)); + assertEval("less", x, y, evaluate("join(rename(x, d0, d2), y, f(a,b)(a < b))", x, y)); + assertEval("equal", x, y, evaluate("join(rename(x, d0, d2), y, f(a,b)(a == b))", x, y)); + assertEval("pow", x, y, evaluate("join(rename(x, d0, d2), y, f(a,b)(pow(a,b)))", x, y)); } @Test @@ -185,9 +197,55 @@ public class OnnxOperationsTestCase { @Test public void testMatMul1() throws ParseException { - Tensor a = evaluate("tensor(d0[2],d1[3]):[1, 2, 3, 4, 5, 6]"); - Tensor b = evaluate("tensor(d0[3],d1[2]):[7, 8, 9, 10, 11, 12]"); - assertEval("matmul", a, b, evaluate("tensor(d0[2],d1[2]):[58, 64, 139, 154]")); + Tensor a = evaluate("tensor(d0[6]):[1,2,3,4,5,6]"); + Tensor b = evaluate("tensor(d0[6]):[1,2,3,4,5,6]"); + assertEval("matmul", a, b, evaluate("91")); + + a = evaluate("tensor(d0[3]):[1,2,3]"); + b = evaluate("tensor(d0[3],d1[2]):[1,2,3,4,5,6]"); + assertEval("matmul", a, b, evaluate("tensor(d0[2]):[22, 28]")); + + a = evaluate("tensor(d0[2],d1[3]):[1,2,3,4,5,6]"); + b = evaluate("tensor(d0[3]):[1,2,3]"); + assertEval("matmul", a, b, evaluate("tensor(d0[2]):[14, 32]")); + + a = evaluate("tensor(d0[2],d1[3]):[1,2,3,4,5,6]"); + b = evaluate("tensor(d0[3],d1[2]):[1,2,3,4,5,6]"); + assertEval("matmul", a, b, evaluate("tensor(d0[2],d1[2]):[22,28,49,64]")); + + a = evaluate("tensor(d0[1],d1[2],d2[3]):[1,2,3,4,5,6]"); + b = evaluate("tensor(d0[3],d1[2]):[1,2,3,4,5,6]"); +// assertEval("matmul", a, b, evaluate("tensor(d0[1],d1[2],d2[2]):[22,28,49,64]")); + + a = evaluate("tensor(d0[2],d1[3]):[1,2,3,4,5,6]"); + b = evaluate("tensor(d0[1],d1[3],d2[2]):[1,2,3,4,5,6]"); +// assertEval("matmul", a, b, evaluate("tensor(d0[1],d1[2],d2[2]):[22,28,49,64]")); + + a = evaluate("tensor(d0[1],d1[2],d2[3]):[1,2,3,4,5,6]"); + b = evaluate("tensor(d0[1],d1[3],d2[2]):[1,2,3,4,5,6]"); + assertEval("matmul", a, b, evaluate("tensor(d0[1],d1[2],d2[2]):[22,28,49,64]")); + + a = evaluate("tensor(d0[1],d1[1],d2[2],d3[3]):[1,2,3,4,5,6]"); + b = evaluate("tensor(d0[1],d1[4],d2[3],d3[2]):[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24]"); +// assertEval("matmul", a, b, evaluate("tensor(d0[1],d1[4],d2[2],d3[2]):[22,28,49,64,58,64,139,154,94,100,229,244,130,136,319,334]")); + + a = evaluate("tensor(d0[1],d1[4],d2[2],d3[3]):[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24]"); + b = evaluate("tensor(d0[1],d1[4],d2[3],d3[2]):[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24]"); + assertEval("matmul", a, b, evaluate("tensor(d0[1],d1[4],d2[2],d3[2]):[22,28,49,64,220,244,301,334,634,676,769,820,1264,1324,1453,1522]")); + + +// a = evaluate("tensor(d0[1],d1[1],d2[2],d3[3]):[1,2,3,4,5,6]"); +// b = evaluate("tensor(d0[1],d1[4],d2[3],d3[2]):[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24]"); +// assertEval("matmul", a, b, evaluate("tensor(d0[1],d1[4],d2[2],d3[2]):[22,28,49,64,58,64,139,154,94,100,229,244,130,136,319,334]")); + +// a = evaluate("tensor(d0[1],d1[2],d2[3]):[1,2,3,4,5,6]"); +// assertEval("matmul", a, b, evaluate("tensor(d0[1],d1[4],d2[2],d3[2]):[22,28,49,64,58,64,139,154,22,28,49,64,58,64,139,154]")); + +// a = evaluate("tensor(d0[2],d1[3]):[1,2,3,4,5,6]"); +// assertEval("matmul", a, b, evaluate("tensor(d0[1],d1[4],d2[2],d3[2]):[22,28,49,64,58,64,139,154,22,28,49,64,58,64,139,154]")); + +// a = evaluate("tensor(d0[3]]):[1,2,3]"); +// assertEval("matmul", a, b, evaluate("tensor(d0[1],d1[4],d2[2]):[22,28,58,64,22,28,58,64]")); } @Test @@ -217,6 +275,10 @@ public class OnnxOperationsTestCase { y = evaluate("tensor(d0[4]):[3,2,-1,1]"); assertEval("reshape", x, y, evaluate("tensor(d0[3],d1[2],d2[1],d3[1]):[1,2,3,4,5,6]")); + + x = evaluate("tensor(d0[1],d1[2],d2[2],d3[3]):[1,2,3,4,5,6,7,8,9,10,11,12]"); + y = evaluate("tensor(d0[2]):[2,6]"); + assertEval("reshape", x, y, evaluate("tensor(d0[2],d1[6]):[1,2,3,4,5,6,7,8,9,10,11,12]")); } @Test @@ -435,6 +497,50 @@ public class OnnxOperationsTestCase { } + @Test + public void testTranspose1() throws ParseException { + Tensor x = evaluate("tensor(d0[2],d1[3]):[[1,2,3],[4,5,6]]"); + assertEval("transpose", x, evaluate("tensor(d0[3],d1[2]):[[1,4],[2,5],[3,6]]")); + } + + @Test + public void testTile6() throws ParseException { + Tensor x = evaluate("tensor(d0[2],d1[2]):[1,2,3,4]"); + Tensor y = evaluate("tensor(d0[2]):[1,2]"); + assertEval("tile", x, y, evaluate("tensor(d0[2],d1[4]):[1,2,1,2,3,4,3,4]")); + + x = evaluate("tensor(d0[2],d1[2]):[1,2,3,4]"); + y = evaluate("tensor(d0[2]):[3,1]"); + assertEval("tile", x, y, evaluate("tensor(d0[6],d1[2]):[1,2,3,4,1,2,3,4,1,2,3,4]")); + + x = evaluate("tensor(d0[1],d1[1],d2[1]):[1]"); + y = evaluate("tensor(d0[3]):[1,6,1]"); + assertEval("tile", x, y, evaluate("tensor(d0[1],d1[6],d2[1]):[1,1,1,1,1,1]")); + + } + + + @Test + public void testSplit2() throws ParseException { + Tensor x = evaluate("tensor(d0[6]):[1,2,3,4,5,6]"); + assertEval("split", x, evaluate("tensor(d0[3]):[1,2,3]"), 0); + assertEval("split", x, evaluate("tensor(d0[3]):[4,5,6]"), 1); + assertEval("split", x, evaluate("tensor(d0[2]):[1,2]"), createAttribute("split", new int[] {2}), 0); + assertEval("split", x, evaluate("tensor(d0[4]):[3,4,5,6]"), createAttribute("split", new int[] {2}), 1); + assertEval("split", x, evaluate("tensor(d0[3]):[3,4,5]"), createAttribute("split", new int[] {2,3}), 1); + assertEval("split", x, evaluate("tensor(d0[1]):[6]"), createAttribute("split", new int[] {2,3}), 2); + + x = evaluate("tensor(d0[2],d1[3]):[1,2,3,4,5,6]"); + assertEval("split", x, evaluate("tensor(d0[1],d1[3]):[1,2,3]")); + assertEval("split", x, evaluate("tensor(d0[1],d1[3]):[1,2,3]"), 0); + assertEval("split", x, evaluate("tensor(d0[1],d1[3]):[4,5,6]"), 1); + assertEval("split", x, evaluate("tensor(d0[1],d1[3]):[1,2,3]"), createAttribute("split", new int[] {1}), 0); + assertEval("split", x, evaluate("tensor(d0[1],d1[3]):[4,5,6]"), createAttribute("split", new int[] {1}), 1); + assertEval("split", x, evaluate("tensor(d0[2],d1[1]):[1,4]"), createAttribute("axis", 1), 0); + assertEval("split", x, evaluate("tensor(d0[2],d1[1]):[2,5]"), createAttribute("axis", 1), 1); + assertEval("split", x, evaluate("tensor(d0[2],d1[1]):[3,6]"), createAttribute("axis", 1), 2); + } + private Tensor evaluate(String expr) throws ParseException { return evaluate(expr, null, null, null); } @@ -461,41 +567,49 @@ public class OnnxOperationsTestCase { } private void assertEval(String opName, Tensor x, Tensor expected) { - assertEval(opName, x, null, null, null, null, expected, null); + assertEval(opName, x, null, null, null, null, expected, null, 0); + } + + private void assertEval(String opName, Tensor x, Tensor expected, int output) { + assertEval(opName, x, null, null, null, null, expected, null, output); } private void assertEval(String opName, Tensor x, Tensor expected, AttributeConverter attr) { - assertEval(opName, x, null, null, null, null, expected, attr); + assertEval(opName, x, null, null, null, null, expected, attr, 0); + } + + private void assertEval(String opName, Tensor x, Tensor expected, AttributeConverter attr, int output) { + assertEval(opName, x, null, null, null, null, expected, attr, output); } private void assertEval(String opName, Tensor x, Tensor y, Tensor expected, AttributeConverter attr) { - assertEval(opName, x, y, null, null, null, expected, attr); + assertEval(opName, x, y, null, null, null, expected, attr, 0); } private void assertEval(String opName, Tensor x, Tensor y, Tensor expected) { - assertEval(opName, x, y, null, null, null, expected, null); + assertEval(opName, x, y, null, null, null, expected, null, 0); } private void assertEval(String opName, Tensor x, Tensor y, Tensor z, Tensor expected) { - assertEval(opName, x, y, z, null, null, expected, null); + assertEval(opName, x, y, z, null, null, expected, null, 0); } private void assertEval(String opName, Tensor x, Tensor y, Tensor z, Tensor expected, AttributeConverter attr) { - assertEval(opName, x, y, z, null, null, expected, attr); + assertEval(opName, x, y, z, null, null, expected, attr, 0); } private void assertEval(String opName, Tensor x, Tensor y, Tensor z, Tensor q, Tensor expected) { - assertEval(opName, x, y, z, q, null, expected, null); + assertEval(opName, x, y, z, q, null, expected, null, 0); } private void assertEval(String opName, Tensor x, Tensor y, Tensor z, Tensor q, Tensor r, Tensor expected) { - assertEval(opName, x, y, z, q, r, expected, null); + assertEval(opName, x, y, z, q, r, expected, null, 0); } - private void assertEval(String opName, Tensor x, Tensor y, Tensor z, Tensor q, Tensor r, Tensor expected, AttributeConverter attr) { + private void assertEval(String opName, Tensor x, Tensor y, Tensor z, Tensor q, Tensor r, Tensor expected, AttributeConverter attr, int output) { Context context = new MapContext(DoubleValue.NaN); List<IntermediateOperation> inputs = createInputs(context, x, y, z, q, r); - IntermediateOperation op = mapOperation(opName, inputs, modelName, opName, attr != null ? attr : createAttributes().build()); + IntermediateOperation op = mapOperation(opName, inputs, modelName, opName, attr != null ? attr : createAttributes().build(), output); optimizeAndRename(opName, op); Tensor result = evaluate(op); assertEquals(expected, result); diff --git a/model-integration/src/test/java/ai/vespa/rankingexpression/importer/onnx/SimpleImportTestCase.java b/model-integration/src/test/java/ai/vespa/rankingexpression/importer/onnx/SimpleImportTestCase.java index 9631bddd93d..abecf4f5cb4 100644 --- a/model-integration/src/test/java/ai/vespa/rankingexpression/importer/onnx/SimpleImportTestCase.java +++ b/model-integration/src/test/java/ai/vespa/rankingexpression/importer/onnx/SimpleImportTestCase.java @@ -35,6 +35,15 @@ public class SimpleImportTestCase { } @Test + public void testConstant() { + ImportedModel model = new OnnxImporter().importModel("test", "src/test/models/onnx/simple/const.onnx"); + + MapContext context = new MapContext(); + Tensor result = model.expressions().get("output").evaluate(context).asTensor(); + assertEquals(result, Tensor.from("tensor():0.42")); + } + + @Test public void testGather() { ImportedModel model = new OnnxImporter().importModel("test", "src/test/models/onnx/simple/gather.onnx"); @@ -48,6 +57,19 @@ public class SimpleImportTestCase { assertEquals(result, Tensor.from("tensor(d0[2],d1[2],d2[2]):[1, 2, 3, 4, 3, 4, 5, 6]")); } + @Test + public void testConcat() { + ImportedModel model = new OnnxImporter().importModel("test", "src/test/models/onnx/simple/concat.onnx"); + + MapContext context = new MapContext(); + context.put("i", new TensorValue(Tensor.from("tensor(d0[1]):[1]"))); + context.put("j", new TensorValue(Tensor.from("tensor(d0[1]):[2]"))); + context.put("k", new TensorValue(Tensor.from("tensor(d0[1]):[3]"))); + + Tensor result = model.expressions().get("y").evaluate(context).asTensor(); + assertEquals(result, Tensor.from("tensor(d0[3]):[1, 2, 3]")); + } + private void evaluateFunction(Context context, ImportedModel model, String functionName) { if (!context.names().contains(functionName)) { RankingExpression e = RankingExpression.from(model.functions().get(functionName)); diff --git a/model-integration/src/test/java/ai/vespa/rankingexpression/importer/tensorflow/LesterTensorflowImportTestCase.java b/model-integration/src/test/java/ai/vespa/rankingexpression/importer/tensorflow/LesterTensorflowImportTestCase.java new file mode 100644 index 00000000000..66af131aa36 --- /dev/null +++ b/model-integration/src/test/java/ai/vespa/rankingexpression/importer/tensorflow/LesterTensorflowImportTestCase.java @@ -0,0 +1,162 @@ +// Copyright 2018 Yahoo Holdings. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. +package ai.vespa.rankingexpression.importer.tensorflow; + +import ai.vespa.rankingexpression.importer.ImportedModel; +import ai.vespa.rankingexpression.importer.configmodelview.ImportedMlFunction; +import ai.vespa.rankingexpression.importer.onnx.OnnxImporter; +import com.yahoo.collections.Pair; +import com.yahoo.searchlib.rankingexpression.RankingExpression; +import com.yahoo.searchlib.rankingexpression.evaluation.Context; +import com.yahoo.searchlib.rankingexpression.evaluation.ContextIndex; +import com.yahoo.searchlib.rankingexpression.evaluation.ExpressionOptimizer; +import com.yahoo.searchlib.rankingexpression.evaluation.MapContext; +import com.yahoo.searchlib.rankingexpression.evaluation.TensorValue; +import com.yahoo.searchlib.rankingexpression.rule.CompositeNode; +import com.yahoo.searchlib.rankingexpression.rule.ExpressionNode; +import com.yahoo.searchlib.rankingexpression.rule.ReferenceNode; +import com.yahoo.system.ProcessExecuter; +import com.yahoo.tensor.Tensor; +import com.yahoo.tensor.TensorType; +import org.junit.Assert; +import org.junit.Ignore; +import org.junit.Test; +import org.tensorflow.SavedModelBundle; +import org.tensorflow.Session; + +import java.io.IOException; +import java.nio.DoubleBuffer; +import java.nio.FloatBuffer; +import java.util.List; +import java.util.Map; + +import static org.junit.Assert.assertEquals; +import static org.junit.Assert.assertNotNull; +import static org.junit.Assert.assertTrue; + +public class LesterTensorflowImportTestCase { + + @Test + @Ignore + public void testPyTorchExport() { + ImportedModel model = new OnnxImporter().importModel("test", "src/test/models/pytorch/test.onnx"); + Tensor onnxResult = evaluateVespa(model, "output", model.inputs()); + assertEquals(Tensor.from("tensor(d0[1],d1[2]):[[0.2134095202835272, -0.08556838456161658]]"), onnxResult); + } + + @Test + @Ignore + public void testBERT() { + ImportedModel model = new OnnxImporter().importModel("test", "src/test/models/onnx/bert/bertsquad10.onnx"); + } + + private Tensor evaluateVespa(ImportedModel model, String operationName, Map<String, TensorType> inputs) { + Context context = contextFrom(model); + for (Map.Entry<String, TensorType> entry : inputs.entrySet()) { + Tensor argument = vespaInputArgument(1, entry.getValue().dimensions().get(1).size().get().intValue()); + context.put(entry.getKey(), new TensorValue(argument)); + } + model.functions().forEach((k, v) -> evaluateFunction(context, model, k)); + RankingExpression expression = model.expressions().get(operationName); + ExpressionOptimizer optimizer = new ExpressionOptimizer(); + optimizer.optimize(expression, (ContextIndex)context); + return expression.evaluate(context).asTensor(); + } + + @Test + @Ignore + public void testModelImport() { + + // MÃ¥ endre til tf 2.0 i java! + + String modelDir = "src/test/models/tensorflow/tf2/saved_model/"; + // output function + String operationName = "out"; + + // Import TF + SavedModelBundle tensorFlowModel = SavedModelBundle.load(modelDir, "serve"); + ImportedModel model = new TensorFlowImporter().importModel("test", modelDir, tensorFlowModel); + ImportedModel.Signature signature = model.signature("serving_default"); + assertEquals("Should have no skipped outputs", 0, model.signature("serving_default").skippedOutputs().size()); + ImportedMlFunction output = signature.outputFunction("output", operationName); + assertNotNull(output); + + // Test TF + Session.Runner runner = tensorFlowModel.session().runner(); + runner.feed("x", tensorFlowFloatInputArgument(1, 4)); + List<org.tensorflow.Tensor<?>> results = runner.fetch(operationName).run(); + assertEquals(1, results.size()); + Tensor tfResult = TensorConverter.toVespaTensor(results.get(0)); + + // Test Vespa + Context context = contextFrom(model); + context.put("x", new TensorValue(vespaInputArgument(1, 4))); + model.functions().forEach((k, v) -> evaluateFunction(context, model, k)); + RankingExpression expression = model.expressions().get(operationName); + ExpressionOptimizer optimizer = new ExpressionOptimizer(); + optimizer.optimize(expression, (ContextIndex)context); + Tensor vespaResult = expression.evaluate(context).asTensor(); + + // Equal result? + System.out.println(tfResult); + System.out.println(vespaResult); + assertEquals(tfResult, vespaResult); + } + + private org.tensorflow.Tensor<?> tensorFlowFloatInputArgument(int d0Size, int d1Size) { + FloatBuffer fb1 = FloatBuffer.allocate(d0Size * d1Size); + int i = 0; + for (int d0 = 0; d0 < d0Size; d0++) + for (int d1 = 0; d1 < d1Size; ++d1) + fb1.put(i++, (float)(d1 * 1.0 / d1Size)); + return org.tensorflow.Tensor.create(new long[]{ d0Size, d1Size }, fb1); + } + + private Tensor vespaInputArgument(int d0Size, int d1Size) { + Tensor.Builder b = Tensor.Builder.of(new TensorType.Builder().indexed("d0", d0Size).indexed("d1", d1Size).build()); + for (int d0 = 0; d0 < d0Size; d0++) + for (int d1 = 0; d1 < d1Size; d1++) + b.cell(d1 * 1.0 / d1Size, d0, d1); + return b.build(); + } + + static Context contextFrom(ImportedModel result) { + TestableModelContext context = new TestableModelContext(); + result.largeConstants().forEach((name, tensor) -> context.put("constant(" + name + ")", new TensorValue(Tensor.from(tensor)))); + result.smallConstants().forEach((name, tensor) -> context.put("constant(" + name + ")", new TensorValue(Tensor.from(tensor)))); + return context; + } + + private void evaluateFunction(Context context, ImportedModel model, String functionName) { + if (!context.names().contains(functionName)) { + RankingExpression e = RankingExpression.from(model.functions().get(functionName)); + evaluateFunctionDependencies(context, model, e.getRoot()); + context.put(functionName, new TensorValue(e.evaluate(context).asTensor())); + } + } + + private void evaluateFunctionDependencies(Context context, ImportedModel model, ExpressionNode node) { + if (node instanceof ReferenceNode) { + String name = node.toString(); + if (model.functions().containsKey(name)) { + evaluateFunction(context, model, name); + } + } + else if (node instanceof CompositeNode) { + for (ExpressionNode child : ((CompositeNode)node).children()) { + evaluateFunctionDependencies(context, model, child); + } + } + } + + private static class TestableModelContext extends MapContext implements ContextIndex { + @Override + public int size() { + return bindings().size(); + } + @Override + public int getIndex(String name) { + throw new UnsupportedOperationException(this + " does not support index lookup by name"); + } + } + +} diff --git a/model-integration/src/test/models/onnx/simple/concat.onnx b/model-integration/src/test/models/onnx/simple/concat.onnx Binary files differnew file mode 100644 index 00000000000..945bc3c9445 --- /dev/null +++ b/model-integration/src/test/models/onnx/simple/concat.onnx diff --git a/model-integration/src/test/models/onnx/simple/concat.py b/model-integration/src/test/models/onnx/simple/concat.py new file mode 100755 index 00000000000..b77cf5decc1 --- /dev/null +++ b/model-integration/src/test/models/onnx/simple/concat.py @@ -0,0 +1,25 @@ +# Copyright 2020 Oath Inc. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. +import onnx +import numpy as np +from onnx import helper, TensorProto + +i_type = helper.make_tensor_value_info('i', TensorProto.FLOAT, [1]) +j_type = helper.make_tensor_value_info('j', TensorProto.FLOAT, [1]) +k_type = helper.make_tensor_value_info('k', TensorProto.FLOAT, [1]) + +output_type = helper.make_tensor_value_info('y', TensorProto.FLOAT, [3]) + +node = onnx.helper.make_node( + 'Concat', + inputs=['i', 'j', 'k'], + outputs=['y'], + axis=0, +) +graph_def = onnx.helper.make_graph( + nodes = [node], + name = 'concat_test', + inputs = [i_type, j_type, k_type], + outputs = [output_type] +) +model_def = helper.make_model(graph_def, producer_name='concat.py') +onnx.save(model_def, 'concat.onnx') diff --git a/model-integration/src/test/models/onnx/simple/const.onnx b/model-integration/src/test/models/onnx/simple/const.onnx Binary files differnew file mode 100644 index 00000000000..c75a92ff12c --- /dev/null +++ b/model-integration/src/test/models/onnx/simple/const.onnx diff --git a/model-integration/src/test/models/onnx/simple/const.py b/model-integration/src/test/models/onnx/simple/const.py new file mode 100755 index 00000000000..35d6dee1346 --- /dev/null +++ b/model-integration/src/test/models/onnx/simple/const.py @@ -0,0 +1,26 @@ +# Copyright 2020 Oath Inc. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. +import onnx +import numpy as np +from onnx import helper, TensorProto + +output_type = helper.make_tensor_value_info('y', TensorProto.FLOAT, []) + +node = onnx.helper.make_node( + 'Constant', + inputs=[], + outputs=['y'], + value=onnx.helper.make_tensor( + name='const_tensor', + data_type=onnx.TensorProto.FLOAT, + dims=(), + vals=[0.42] + ), +) +graph_def = onnx.helper.make_graph( + nodes = [node], + name = 'constant_test', + inputs = [], + outputs = [output_type] +) +model_def = helper.make_model(graph_def, producer_name='const.py') +onnx.save(model_def, 'const.onnx') diff --git a/model-integration/src/test/models/onnx/simple/gather.onnx b/model-integration/src/test/models/onnx/simple/gather.onnx Binary files differindex 62451ad953d..0647d86ed0f 100644 --- a/model-integration/src/test/models/onnx/simple/gather.onnx +++ b/model-integration/src/test/models/onnx/simple/gather.onnx diff --git a/model-integration/src/test/models/onnx/simple/simple.onnx b/model-integration/src/test/models/onnx/simple/simple.onnx index 1c746c90efa..41b458451d0 100644 --- a/model-integration/src/test/models/onnx/simple/simple.onnx +++ b/model-integration/src/test/models/onnx/simple/simple.onnx @@ -1,4 +1,4 @@ - simple.py:ã + simple.py:ã 0 query_tensor attribute_tensormatmul"MatMul @@ -20,4 +20,4 @@ output -B +B
\ No newline at end of file |