Safetensors
whisper
huseinzol05 commited on
Commit
2b29532
·
verified ·
1 Parent(s): 8ae08d9

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +16 -17
README.md CHANGED
@@ -21,6 +21,21 @@ Finetune Whisper Large V3 Turbo on Malaysian context.
21
  2. Better translation for Malay, Manglish, Mandarin, Tamil and Science context.
22
  3. Word level timestamp, introduced `<|transcribeprecise|>` token, **a new task!**
23
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
24
  ## Example
25
 
26
  We use for examples,
@@ -456,20 +471,4 @@ Output,
456
  <|28.96|> penat.<|29.32|>
457
  <|29.36|> Kelangkabut<|29.72|>
458
  <|endoftext|>
459
- ```
460
-
461
- ## how we finetuned it?
462
-
463
- We done 2 phases,
464
-
465
- 1. Finetune on [mesolitica/Malaysian-STT-Whisper](https://huggingface.co/datasets/mesolitica/Malaysian-STT-Whisper)
466
- - Revision [267552e0f093068519a816112c2741939d057f48](https://huggingface.co/mesolitica/malaysian-whisper-large-v3-turbo-v3/commit/267552e0f093068519a816112c2741939d057f48)
467
- - WanDB at https://wandb.ai/huseinzol05/malaysian-whisper-large-v3-turbo-v3
468
- 2. Stage 2, Annealing on 5% from [mesolitica/Malaysian-STT-Whisper](https://huggingface.co/datasets/mesolitica/Malaysian-STT-Whisper) and 100% from [mesolitica/Malaysian-STT-Whisper-Stage2](https://huggingface.co/datasets/mesolitica/Malaysian-STT-Whisper-Stage2)
469
- - Revision [5f6ca0596f01527ec2d013662cfc168d5f754461](https://huggingface.co/mesolitica/malaysian-whisper-large-v3-turbo-v3/commit/5f6ca0596f01527ec2d013662cfc168d5f754461)
470
- - WanDB at https://wandb.ai/huseinzol05/malaysian-whisper-large-v3-turbo-v3-stage2
471
-
472
- ## Benchmark
473
-
474
-
475
- ![image/png](https://cdn-uploads.huggingface.co/production/uploads/5e73316106936008a9ee6523/1_3mUW5Au9DWcO-PnJ9l8.png)
 
21
  2. Better translation for Malay, Manglish, Mandarin, Tamil and Science context.
22
  3. Word level timestamp, introduced `<|transcribeprecise|>` token, **a new task!**
23
 
24
+ ## how we finetuned it?
25
+
26
+ We done 2 phases,
27
+
28
+ 1. Finetune on [mesolitica/Malaysian-STT-Whisper](https://huggingface.co/datasets/mesolitica/Malaysian-STT-Whisper)
29
+ - Revision [267552e0f093068519a816112c2741939d057f48](https://huggingface.co/mesolitica/malaysian-whisper-large-v3-turbo-v3/commit/267552e0f093068519a816112c2741939d057f48)
30
+ - WanDB at https://wandb.ai/huseinzol05/malaysian-whisper-large-v3-turbo-v3
31
+ 2. Stage 2, Annealing on 5% from [mesolitica/Malaysian-STT-Whisper](https://huggingface.co/datasets/mesolitica/Malaysian-STT-Whisper) and 100% from [mesolitica/Malaysian-STT-Whisper-Stage2](https://huggingface.co/datasets/mesolitica/Malaysian-STT-Whisper-Stage2)
32
+ - Revision [5f6ca0596f01527ec2d013662cfc168d5f754461](https://huggingface.co/mesolitica/malaysian-whisper-large-v3-turbo-v3/commit/5f6ca0596f01527ec2d013662cfc168d5f754461)
33
+ - WanDB at https://wandb.ai/huseinzol05/malaysian-whisper-large-v3-turbo-v3-stage2
34
+
35
+ ## Benchmark
36
+
37
+ ![image/png](https://cdn-uploads.huggingface.co/production/uploads/5e73316106936008a9ee6523/1_3mUW5Au9DWcO-PnJ9l8.png)
38
+
39
  ## Example
40
 
41
  We use for examples,
 
471
  <|28.96|> penat.<|29.32|>
472
  <|29.36|> Kelangkabut<|29.72|>
473
  <|endoftext|>
474
+ ```