summaryrefslogtreecommitdiffstats
path: root/configdefinitions
diff options
context:
space:
mode:
Diffstat (limited to 'configdefinitions')
-rw-r--r--configdefinitions/src/vespa/embedding.bert-base-embedder.def9
1 files changed, 6 insertions, 3 deletions
diff --git a/configdefinitions/src/vespa/embedding.bert-base-embedder.def b/configdefinitions/src/vespa/embedding.bert-base-embedder.def
index a37599de411..115e021972c 100644
--- a/configdefinitions/src/vespa/embedding.bert-base-embedder.def
+++ b/configdefinitions/src/vespa/embedding.bert-base-embedder.def
@@ -1,8 +1,13 @@
namespace=embedding
+# Settings for wordpiece tokenizer
+tokenizerVocabUrl url
+tokenizerVocabPath path
+
# Transformer model settings
-transformerModelUrl url default=https://data.vespa.oath.cloud/onnx_models/sentence_all_MiniLM_L6_v2.onnx
+transformerModelUrl url
+transformerModelPath path
# Max length of token sequence model can handle
transformerMaxTokens int default=384
@@ -23,5 +28,3 @@ onnxExecutionMode enum { parallel, sequential } default=sequential
onnxInterOpThreads int default=1
onnxIntraOpThreads int default=-4 # n=number of threads -> n<0: CPUs/(-n), n==0: CPUs, n>0: n
-# Settings for wordpiece tokenizer
-tokenizerVocabUrl url default=https://data.vespa.oath.cloud/onnx_models/bert-base-uncased-vocab.txt