Hugging Face's logo

OpenMOSS-Team
/
MOSS-Audio-8B-Instruct

Audio-Text-to-Text
Safetensors
English
Chinese
moss_audio
audio
speech
music
understanding
multimodal
instruct
custom_code
Model card Files
xet
 Community
3

Fix MOSS-Audio processor config

#3
by kiiic - opened
base: refs/heads/main
←
from: refs/pr/3
Discussion Files changed
+33
-3
Files changed (2) hide show
  1. preprocessor_config.json +15 -3
  2. processor_config.json +18 -0
preprocessor_config.json CHANGED
@@ -1,6 +1,18 @@
1
  {
2
- "processor_class": "PalomarProcessor",
3
  "auto_map": {
4
- "AutoProcessor": "processing_palomar_myaut_mel_whisper.PalomarProcessor"
5
- }
 
 
 
 
 
 
 
 
 
 
 
 
6
  }
 
1
  {
2
+ "processor_class": "MossAudioProcessor",
3
  "auto_map": {
4
+ "AutoProcessor": "processing_moss_audio.MossAudioProcessor"
5
+ },
6
+ "mel_config": {
7
+ "mel_sr": 16000,
8
+ "mel_dim": 128,
9
+ "mel_n_fft": 400,
10
+ "mel_hop_length": 160,
11
+ "mel_dtype": "bfloat16",
12
+ "use_whisper_feature_extractor": true
13
+ },
14
+ "enable_time_marker": true,
15
+ "audio_token_id": 151654,
16
+ "audio_start_id": 151669,
17
+ "audio_end_id": 151670
18
  }
processor_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "processor_class": "MossAudioProcessor",
3
+ "auto_map": {
4
+ "AutoProcessor": "processing_moss_audio.MossAudioProcessor"
5
+ },
6
+ "mel_config": {
7
+ "mel_sr": 16000,
8
+ "mel_dim": 128,
9
+ "mel_n_fft": 400,
10
+ "mel_hop_length": 160,
11
+ "mel_dtype": "bfloat16",
12
+ "use_whisper_feature_extractor": true
13
+ },
14
+ "enable_time_marker": true,
15
+ "audio_token_id": 151654,
16
+ "audio_start_id": 151669,
17
+ "audio_end_id": 151670
18
+ }