ModelPath
The path to the exported model file to use. This parameter applies when Type is set to ExtractiveQuestionAnsweringLLM or GenerativeLLM.
You can generate an appropriate model file to use by using the export_transformers_model.py
script, which is provided in your Answer Server installation.
-
When Type is set to
ExtractiveQuestionAnsweringLLM, you must use an ALBERT model (see https://huggingface.co/docs/transformers/model_doc/albert). -
When Type is set to
GenerativeLLM, you must use a T5 model (see https://huggingface.co/docs/transformers/model_doc/t5).
The script converts these models into a format that Answer Server can use. For more information about how to use the script, see Create the Model Files.
| Type: | String |
| Default: | |
| Required: | Yes, when Type is set to ExtractiveQuestionAnsweringLLM or GenerativeLLM |
| Configuration Section: | MyLLMModule |
| Example: | ModelPath=C:\Models\model.pt
|
| See Also: | TokenizerPath |