Hugging Face's logo

amd-quark
/
llama-small-int4-per-group-sym-awq

Safetensors
llama
quark
Model card Files Community
llama-small-int4-per-group-sym-awq
File size: 2,815 Bytes 
dc62f88
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
 
{
  "_name_or_path": "fxmarty/small-llama-testing",
  "architectures": [
    "LlamaForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "bos_token_id": 0,
  "eos_token_id": 1,
  "head_dim": 64,
  "hidden_act": "silu",
  "hidden_size": 256,
  "initializer_range": 0.02,
  "intermediate_size": 128,
  "max_position_embeddings": 2048,
  "mlp_bias": false,
  "model_type": "llama",
  "num_attention_heads": 4,
  "num_hidden_layers": 2,
  "num_key_value_heads": 4,
  "pad_token_id": -1,
  "pretraining_tp": 1,
  "quantization_config": {
    "algo_config": {
      "model_decoder_layers": "model.layers",
      "name": "awq",
      "scaling_layers": [
        {
          "inp": "self_attn.q_proj",
          "layers": [
            "self_attn.q_proj",
            "self_attn.k_proj",
            "self_attn.v_proj"
          ],
          "module2inspect": "self_attn",
          "prev_op": "input_layernorm"
        },
        {
          "condition": "module.self_attn.v_proj.weight.shape == module.self_attn.o_proj.weight.shape",
          "help": "attention out, Please refer to https://github.com/mit-han-lab/llm-awq/pull/67#issue-1850622696, if module.self_attn.v_proj.weight.shape == module.self_attn.o_proj.weight.shape",
          "inp": "self_attn.o_proj",
          "layers": [
            "self_attn.o_proj"
          ],
          "prev_op": "self_attn.v_proj"
        },
        {
          "inp": "mlp.gate_proj",
          "layers": [
            "mlp.gate_proj",
            "mlp.up_proj"
          ],
          "module2inspect": "mlp",
          "prev_op": "post_attention_layernorm"
        },
        {
          "inp": "mlp.down_proj",
          "layers": [
            "mlp.down_proj"
          ],
          "prev_op": "mlp.up_proj"
        }
      ]
    },
    "exclude": [
      "lm_head"
    ],
    "export": {
      "kv_cache_group": [],
      "pack_method": "reorder",
      "weight_format": "real_quantized",
      "weight_merge_groups": null
    },
    "global_quant_config": {
      "bias": null,
      "input_tensors": null,
      "output_tensors": null,
      "target_device": null,
      "weight": {
        "ch_axis": 1,
        "dtype": "int4",
        "group_size": 4,
        "is_dynamic": false,
        "observer_cls": "PerGroupMinMaxObserver",
        "qscheme": "per_group",
        "round_method": "half_even",
        "scale_type": "float",
        "symmetric": true
      }
    },
    "layer_quant_config": {},
    "layer_type_quant_config": {},
    "pack_method": "reorder",
    "quant_method": "quark",
    "quant_mode": "eager_mode"
  },
  "rms_norm_eps": 1e-06,
  "rope_scaling": null,
  "rope_theta": 10000.0,
  "tie_word_embeddings": false,
  "torch_dtype": "float32",
  "transformers_version": "4.45.2",
  "use_cache": true,
  "vocab_size": 32000
}