Add new model to the arena #2296
Conversation
fastchat/model/model_adapter.py
Outdated
| @@ -1547,6 +1564,7 @@ def get_default_conv_template(self, model_path: str) -> Conversation: | |||
|
|
|||
| # Note: the registration order matters. | |||
| # The one registered earlier has a higher matching priority. | |||
| register_model_adapter(ReaLMAdapter) | |||
There was a problem hiding this comment.
move this to the end of the list
There was a problem hiding this comment.
move this to the end of the list
done it, thx!!
There was a problem hiding this comment.
move this to the end of the list
we also infer and evaluate our model by ourselves, do you need the JSON file?
There was a problem hiding this comment.
do you mean mt-bench JSON files? please share it with a public link and we may add it to our database later.
There was a problem hiding this comment.
do you mean mt-bench JSON files? please share it with a public link and we may add it to our database later.
yeah. it was uploaded to our repo:
https://github.com/FreedomIntelligence/ReaLM/tree/main/data/mt-bench
There was a problem hiding this comment.
do you mean mt-bench JSON files? please share it with a public link and we may add it to our database later.
hi, bro. can I know when will your group add our model to mt-bench?
we achieved the best scores on MT-Bench's 7b level with less samples and shorter context length through a trainable simulator~
|
sorry, sir, my boss wants to rename the model name, can I pr again? the github links: |
hi, thanks for your brilliant job!!
we want to support our model
ReaLMin mt-bench.we have added the related information to
conversation.py,model_adapter.py,model_registry.py,model_support.md.all the information is coded under our previous model
phoenix.the following links are the repos for the code and model.
https://github.com/FreedomIntelligence/ReaLM
https://huggingface.co/FreedomIntelligence/ReaLM-7b
thanks in advance!