Lots of good models use Orca for their merges, however vanilla Orca has vocabulary size of 32003, 3 last tokens are ChatML tokens and a PAD token. This causes errors during a merge with models with standard 32000 vocabulary size.
I've removed those tokens from volabulary and resized model embeddings to mach 32000 standard size. So this model is ready to be used as a merge component in mergekit. It may not work on its own with ChatML template anymore.
model.resize_token_embeddings(32000)
- Downloads last month
- 9
Install from pip and serve model
# Install vLLM from pip: pip install vllm# Start the vLLM server: vllm serve "TeeZee/Orca-2-13b_flat"# Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "TeeZee/Orca-2-13b_flat", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'