Hugging Face's logo

fukayatti0
/
nllb-200-distilled-600M-4bit-efqat

Translation
PyTorch
multilingual
m2m_100
quantization
efqat
nllb
Eval Results
4-bit precision
Model card Files Community
nllb-200-distilled-600M-4bit-efqat
File size: 1,245 Bytes 
050d6d9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
 
{
  "_name_or_path": "facebook/nllb-200-distilled-600M",
  "architectures": [
    "M2M100ForConditionalGeneration"
  ],
  "model_type": "m2m_100",
  "quantization_config": {
    "method": "EfQAT-CWPN",
    "bits": 4,
    "adaptive_bits": true,
    "critical_layers_bits": 8,
    "normal_layers_bits": 4,
    "compression_ratio": 6.3,
    "precision_retention": 0.807
  },
  "efqat_config": {
    "update_ratio": 0.08,
    "update_frequency": 30,
    "progressive_stages": 3,
    "critical_layer_patterns": [
      "self_attn.q_proj",
      "self_attn.k_proj",
      "self_attn.v_proj",
      "encoder_attn.q_proj",
      "encoder_attn.k_proj",
      "encoder_attn.v_proj",
      "lm_head"
    ]
  },
  "torch_dtype": "float32",
  "transformers_version": "4.36.0",
  "use_cache": true,
  "vocab_size": 256206,
  "d_model": 1024,
  "encoder_layers": 12,
  "decoder_layers": 12,
  "encoder_attention_heads": 16,
  "decoder_attention_heads": 16,
  "decoder_ffn_dim": 4096,
  "encoder_ffn_dim": 4096,
  "activation_function": "relu",
  "dropout": 0.1,
  "attention_dropout": 0.1,
  "activation_dropout": 0.0,
  "init_std": 0.02,
  "decoder_start_token_id": 2,
  "scale_embedding": true,
  "pad_token_id": 1,
  "bos_token_id": 0,
  "eos_token_id": 2
}