Update README.md

README.md

@@ -30,7 +30,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # Distil-Whisper Small zh-HK - Alvin
 
-This model is a distilled and fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the Cantonese language. It achieves a 9.77 CER (without punctuations), 11.7 CER (with punctuations) on Common Voice 16.0
+This model is a distilled version of [alvanlii/whisper-small-cantonese](https://huggingface.co/alvanlii/whisper-small-cantonese) on the Cantonese language. It achieves a 9.77 CER (without punctuations), 11.7 CER (with punctuations) on Common Voice 16.0. It has 6 decoder layers instead of 12.
 
 ## Training and evaluation data
 For training,
@@ -41,11 +41,11 @@ For training,
 For evaluation, Common Voice 16.0 yue Test set is used.
 
 ## Results
-- CER (lower is better): 0.117
-- GPU Inference with Fast Attention (example below): 0.039s/sample
+- CER (lower is better): 0.117 (compared to 0.107 for `alvanlii/whisper-small-cantonese`)
+- GPU Inference with Fast Attention (sdpa): 0.039s/sample (down from 0.055s)
   - Note all GPU evaluations are done on RTX 3090 GPU
-- GPU Inference: <TODO>s/sample
-- CPU Inference: 2.57s/sample
+- GPU Inference: 0.041s/sample (down from 0.308s)
+- CPU Inference: 1.7s/sample (down from 2.57s)
 - GPU VRAM: ~2 GB
 
 
@@ -89,17 +89,3 @@ pipe = pipeline(
 pipe.model.config.forced_decoder_ids = pipe.tokenizer.get_decoder_prompt_ids(language=lang, task="transcribe")
 text = pipe(file)["text"]
 ```
-
-## Model Speedup
-Just add attn_implementation="sdpa" for Flash Attention. 
-```
-model = AutoModelForSpeechSeq2Seq.from_pretrained(
-    "alvanlii/distil-whisper-small-cantonese",
-    torch_dtype=torch_dtype,
-    low_cpu_mem_usage=True,
-    use_safetensors=True,
-    attn_implementation="sdpa",
-)
-```
-Using Flash Attention reduced the amount of time taken per sample from <TODO>s to 0.039s.
-