Using OPT-125M as starter model

georgian-io · May 7, 2024 · 293d474 · 293d474
1 parent 647f4e5
commit 293d474
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/llmtune/config.yml b/llmtune/config.yml
@@ -23,7 +23,7 @@ data:
 
 # Model Definition -------------------
 model:
-  hf_model_ckpt: "mistralai/Mistral-7B-Instruct-v0.2"
+  hf_model_ckpt: "facebook/opt-125m"
   torch_dtype: "bfloat16"
   #attn_implementation: "flash_attention_2"
   quantize: true