mispeech
/

midashenglm-7b-0804-bf16

Audio-Text-to-Text

audio-language-model

Model card Files Files and versions

jimbozhang commited on 25 days ago

Commit

1262f3f

·

verified ·

1 Parent(s): f78687c

Update README.md

Files changed (1) hide show

README.md +3 -2

README.md CHANGED Viewed

@@ -20,8 +20,9 @@ base_model_relation: finetune
 The bfloat16 (bf16) weights for [mispeech/midashenglm-7b-0804-fp32](https://huggingface.co/mispeech/midashenglm-7b-0804-fp32).
-For production deployments, use this bf16 version for optimal efficiency. The original fp32 model is for  strict numerical reproduction of benchmark results.
 ## Usage

 The bfloat16 (bf16) weights for [mispeech/midashenglm-7b-0804-fp32](https://huggingface.co/mispeech/midashenglm-7b-0804-fp32).
+Recommended for most general-purpose scenarios, including inference and fine-tuning.
+It delivers quality comparable to FP32 while being significantly faster on modern GPUs (e.g., A100, H100, RTX 4090).
+The [original fp32 model](https://huggingface.co/mispeech/midashenglm-7b-0804-fp32) is only for strict numerical reproduction of benchmark results.
 ## Usage