Issue with uploading a custom model in OpenSearch 2.19

OpenSearch
1

Versions (relevant - OpenSearch/Dashboard/Server OS/Browser): 2.19

Describe the issue:

I am trying to upload a custom ML model following this guide

The model is downloaded from

I have converted the file locally using pytorch

This is wheat I am sending along with register method to my local OS
{
“name”: “bge-base-en-v1.5”,
“description”: “BGE Base English v1.5 model from Hugging Face”,
“version”: “1.0.0”,
“function_name”: “TEXT_EMBEDDING”,
“model_group_id”: “GRP ID”,
“model_format”: “TORCH_SCRIPT”,
“is_local_model”: true,
“zip_base64”: “$BASE64_MODEL”,
“model_content_hash_value”: “$MODEL_HASH”,
“model_config”: {
“framework_type”: “sentence_transformers”,
“model_type”: “text_embedding”,
“embedding_dimension”: 786
}
}

It uploads but when I check task statue I am getting this error

This model is not in the pre-trained model list, please check your parameters

I am going through params, not sure about the issue.

Though I am using register model, not sending url (as recommended in guide) as my model is local converted model, not from an external url.

Configuration**:

I set the following as were suggested in the guide

“persistent”: {
“plugins.ml_commons.allow_registering_model_via_url”: “true”,
“plugins.ml_commons.only_run_on_ml_node”: “false”,
“plugins.ml_commons.model_access_control_enabled”: “true”,
“plugins.ml_commons.native_memory_threshold”: “99”
}

Relevant Logs or Screenshots:

[ERROR][o.o.m.m.MLModelManager ] [c4896af2b68d] Failed to update model group

| java.lang.IllegalArgumentException: This model is not in the pre-trained model list, please check your parameters.

at org.opensearch.ml.model.MLModelManager.registerPrebuiltModel(MLModelManager.java:970) ~[?:?]

at org.opensearch.ml.model.MLModelManager.uploadModel(MLModelManager.java:795) ~[?:?]

at org.opensearch.ml.model.MLModelManager.lambda$registerMLModel$15(MLModelManager.java:517) ~[?:?]

at org.opensearch.core.action.ActionListener$1.onResponse(ActionListener.java:82) [opensearch-core-2.19.0.jar:2.19.0]

at org.opensearch.action.support.TransportAction$1.onResponse(TransportAction.java:115) [opensearch-2.19.0.jar:2.19.0]

at org.opensearch.action.support.TransportAction$1.onResponse(TransportAction.java:109) [opensearch-2.19.0.jar:2.19.0]

at org.opensearch.core.action.ActionListener$1.onResponse(ActionListener.java:82) [opensearch-core-2.19.0.jar:2.19.0]

2

ok I managed find another way for fix it by loading the model locally,
so download onnx file from the huggingface site for the model, added with config and tokenizer file, created zip and uploaded the model using following params

{

   "model_group_id": "LuAi65gBYY2A7cZ1233",

   "name": "bge-base-onnx",

   "version": "1.0.0",

   "model_content_hash_value": "<SHA256>",

   "model_format": "ONNX",

   "model_type": "text_embedding",

   "url": "file:///.......",

   "model_config": {

“framework_type”: “sentence_transformers”,

“model_type”: “text_embedding”,

“embedding_dimension”: 768

} }

This allows the model to register but when I deploy, I am seeing the following issue
Input mismatch, looking for: [input_ids, attention_mask, token_type_ids]\"….