Seungyoun commited on
Commit
0ff5ef0
·
verified ·
1 Parent(s): 481b88e

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +11 -4
README.md CHANGED
@@ -1,4 +1,11 @@
1
- # SoriSpeech-4B-Omni
 
 
 
 
 
 
 
2
 
3
  Speech-to-Text model combining:
4
  - **Audio Encoder**: Qwen3-Omni-30B-A3B-Instruct (pretrained weights)
@@ -20,12 +27,12 @@ import torchaudio
20
 
21
  # Load model
22
  model = SoriSpeechForConditionalGeneration.from_pretrained(
23
- "Seungyoun/SoriSpeechLLM-4B-Omni",
24
  torch_dtype=torch.bfloat16,
25
  device_map="auto",
26
  trust_remote_code=True,
27
  )
28
- tokenizer = AutoTokenizer.from_pretrained("Seungyoun/SoriSpeechLLM-4B-Omni")
29
 
30
  # Process audio
31
  audio, sr = torchaudio.load("speech.wav")
@@ -44,4 +51,4 @@ mel_spec = torch.log(mel_transform(audio) + 1e-6)
44
 
45
  ## License
46
 
47
- Apache 2.0 (following Qwen3-Omni and Qwen3 licenses)
 
1
+ ---
2
+ language:
3
+ - ko
4
+ - en
5
+ base_model:
6
+ - Qwen/Qwen3-4B-Instruct-2507
7
+ ---
8
+ # Sori-4B
9
 
10
  Speech-to-Text model combining:
11
  - **Audio Encoder**: Qwen3-Omni-30B-A3B-Instruct (pretrained weights)
 
27
 
28
  # Load model
29
  model = SoriSpeechForConditionalGeneration.from_pretrained(
30
+ "Seungyoun/Sori-4B",
31
  torch_dtype=torch.bfloat16,
32
  device_map="auto",
33
  trust_remote_code=True,
34
  )
35
+ tokenizer = AutoTokenizer.from_pretrained("Seungyoun/Sori-4B")
36
 
37
  # Process audio
38
  audio, sr = torchaudio.load("speech.wav")
 
51
 
52
  ## License
53
 
54
+ Apache 2.0 (following Qwen3-Omni and Qwen3 licenses)