tune the threshold of gemma-2-27b-it in test_nightly_gsm8k_eval.py (#5677)
This commit is contained in:
@@ -24,7 +24,7 @@ MODEL_SCORE_THRESHOLDS = {
|
|||||||
"meta-llama/Llama-3.1-8B-Instruct": 0.82,
|
"meta-llama/Llama-3.1-8B-Instruct": 0.82,
|
||||||
"mistralai/Mistral-7B-Instruct-v0.3": 0.58,
|
"mistralai/Mistral-7B-Instruct-v0.3": 0.58,
|
||||||
"deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct": 0.85,
|
"deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct": 0.85,
|
||||||
"google/gemma-2-27b-it": 0.92,
|
"google/gemma-2-27b-it": 0.91,
|
||||||
"meta-llama/Llama-3.1-70B-Instruct": 0.95,
|
"meta-llama/Llama-3.1-70B-Instruct": 0.95,
|
||||||
"mistralai/Mixtral-8x7B-Instruct-v0.1": 0.64,
|
"mistralai/Mixtral-8x7B-Instruct-v0.1": 0.64,
|
||||||
"Qwen/Qwen2-57B-A14B-Instruct": 0.86,
|
"Qwen/Qwen2-57B-A14B-Instruct": 0.86,
|
||||||
|
|||||||
Reference in New Issue
Block a user