Update lora readme.md, update mistral-7b-v0.1 to mistral-7b-v0.2-instruct

This commit is contained in:
madroid
2023-12-25 22:11:46 +08:00
parent 647e48870a
commit c5cb83b9a0
2 changed files with 4 additions and 2 deletions

View File

@@ -31,8 +31,8 @@ pip install -r requirements.txt
Next, download and convert the model. The Mistral weights can be downloaded with:
```
curl -O https://files.mistral-7b-v0-1.mistral.ai/mistral-7B-v0.1.tar
tar -xf mistral-7B-v0.1.tar
curl -O https://files.mistral-7b-v0-2.mistral.ai/Mistral-7B-v0.2-Instruct.tar
tar -xf Mistral-7B-v0.2-Instruct.tar
```
If you do not have access to the Llama weights you will need to [request

View File

@@ -57,6 +57,8 @@ if __name__ == "__main__":
if k in config:
config.pop(k)
n_heads = config["n_heads"]
if "rope_theta" in config:
config.pop("rope_theta")
if "sliding_window" in config:
config.pop("sliding_window")
if "n_kv_heads" not in config: