AlaBoussoffara commited on
Commit
48ec60d
·
verified ·
1 Parent(s): 00d0cc1

Upload self-contained SmoothQuant W8A8 bundle

Browse files
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
README.txt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ Self-contained W8A8 SmoothQuant bundle.
2
+ No base checkpoint is required to load this bundle.
3
+ Use the loader cell in this notebook to reconstruct model modules and load state_dict.
benchmark_loader_manifest.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "artifact_type": "compressingvlm_w8a8_smoothquant_bundle",
3
+ "format_version": 1,
4
+ "quant_state_file": "w8a8_smoothquant_state.pt",
5
+ "quant_meta_file": "w8a8_smoothquant_meta.json",
6
+ "self_contained_bundle": true,
7
+ "requires_base_model_download": false
8
+ }
chat_template.jinja ADDED
@@ -0,0 +1,138 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- set image_placeholder = '<|vision_start|><|image_pad|><|vision_end|>' -%}
2
+ {%- for message in messages -%}
3
+ {%- if message['role'] == 'system' -%}
4
+ {{- '<|im_start|>system
5
+ ' -}}
6
+ {%- if message['content'] is string -%}
7
+ {{- message['content'] | trim -}}
8
+ {%- endif -%}
9
+ {{- '<|im_end|>
10
+ ' -}}
11
+ {%- elif message['role'] == 'user' -%}
12
+ {%- if loop.first -%}
13
+ {{- '<|im_start|>system
14
+ ' -}}
15
+ {%- if template -%}
16
+ {#--- If template, extraction task ---#}
17
+ {{- 'You are NuExtract, an information extraction tool created by NuMind.' -}}
18
+ {%- else -%}
19
+ {#--- Else, template generation task ---#}
20
+ {{- 'You are a helpful assistant.' -}}
21
+ {%- endif -%}
22
+ {{ '<|im_end|>
23
+ ' }}
24
+ {%- endif -%}
25
+ {{- '<|im_start|>' + message['role'] + '
26
+ ' -}}
27
+ {%- if template -%}
28
+ {#--- Template Section ---#}
29
+ {{- '# Template:
30
+ ' -}}
31
+ {{- template -}}
32
+ {{- '
33
+ ' -}}
34
+
35
+ {%- if examples -%}
36
+ {#--- Examples can only exist in the extraction task ---#}
37
+ {{- '# Examples:
38
+ ' -}}
39
+ {%- for example in examples -%}
40
+ {{- '## Input:
41
+ ' -}}
42
+ {%- if example['input'] is mapping and (example['input']['type'] == 'image' or example['input']['type'] == 'image_url') -%}
43
+ {{- image_placeholder | trim -}}
44
+ {%- elif example['input'] == '<image>' -%}
45
+ {#--- Keep compatibility with <image> for now ---#}
46
+ {{- image_placeholder | trim -}}
47
+ {%- else -%}
48
+ {#--- Text input example ---#}
49
+ {{- example['input'] -}}
50
+ {%- endif -%}
51
+ {{- '
52
+ ' -}}
53
+ {{- '## Output:
54
+ ' -}}
55
+ {{- example['output'] -}}
56
+ {{- '
57
+ ' -}}
58
+ {%- endfor -%}
59
+ {%- endif -%}
60
+ {{- '# Context:
61
+ ' -}}
62
+ {%- endif -%}
63
+
64
+ {%- if message['content'] is string -%}
65
+ {#--- Simple string content ---#}
66
+ {{- message['content'] | trim -}}
67
+ {%- elif message['content'] is mapping and (message['content']['type'] == 'image' or message['content']['type'] == 'image_url') -%}
68
+ {{- image_placeholder | trim -}}
69
+ {%- else -%}
70
+ {#--- List of content items (mixed text/images) ---#}
71
+ {#--- First, determine what the actual input content is (not ICL images) ---#}
72
+ {%- set ns = namespace(has_text_input=false, text_content='') -%}
73
+
74
+ {#--- Count content types and identify actual input document ---#}
75
+ {%- for content in message['content'] -%}
76
+ {%- if content is mapping and content.get('type') == 'text' -%}
77
+ {%- if content.get('text') != '<image>' -%}
78
+ {#--- Keep compatibility with <image> for now ---#}
79
+ {%- set ns.has_text_input = true -%}
80
+ {%- set ns.text_content = content['text'] -%}
81
+ {%- endif -%}
82
+ {%- elif content is string -%}
83
+ {%- if content != '<image>' -%}
84
+ {#--- Keep compatibility with <image> for now ---#}
85
+ {%- set ns.has_text_input = true -%}
86
+ {%- set ns.text_content = content -%}
87
+ {%- endif -%}
88
+ {%- endif -%}
89
+ {%- endfor -%}
90
+
91
+ {#--- Determine what to output based on actual input type ---#}
92
+ {%- if ns.has_text_input -%}
93
+ {#--- Main input is text, so output the text content ---#}
94
+ {{- ns.text_content | trim -}}
95
+ {%- else -%}
96
+ {#--- Main input is image or <image> placeholder ---#}
97
+ {%- set ns2 = namespace(found_image=false) -%}
98
+ {%- for content in message['content'] -%}
99
+ {%- if content is mapping and (content.get('type') == 'image' or content.get('type') == 'image_url') and not ns2.found_image -%}
100
+ {{- image_placeholder | trim -}}
101
+ {%- set ns2.found_image = true -%}
102
+ {%- elif content is mapping and content.get('type') == 'text' and content.get('text') == '<image>' and not ns2.found_image -%}
103
+ {#--- Keep compatibility with <image> for now ---#}
104
+ {{- image_placeholder | trim -}}
105
+ {%- set ns2.found_image = true -%}
106
+ {%- elif content is string and content == '<image>' and not ns2.found_image -%}
107
+ {#--- Keep compatibility with <image> for now ---#}
108
+ {{- image_placeholder | trim -}}
109
+ {%- set ns2.found_image = true -%}
110
+ {%- endif -%}
111
+ {%- endfor -%}
112
+ {%- endif -%}
113
+ {%- endif -%}
114
+ {{- '<|im_end|>
115
+ '}}
116
+
117
+ {%- elif message['role'] == 'assistant' -%}
118
+ {{- '<|im_start|>assistant
119
+ ' -}}
120
+ {%- if message['content'] is string -%}
121
+ {{- message['content'] | trim -}}
122
+ {%- elif message['content'] is iterable and message['content'] is not string -%}
123
+ {%- for content in message['content'] -%}
124
+ {%- if content is mapping and content.get('type') == 'text' -%}
125
+ {{- content['text'] | trim -}}
126
+ {%- elif content is string -%}
127
+ {{- content | trim -}}
128
+ {%- endif -%}
129
+ {%- endfor -%}
130
+ {%- endif -%}
131
+ {{- '<|im_end|>
132
+ ' -}}
133
+ {%- endif -%}
134
+ {%- endfor -%}
135
+ {%- if add_generation_prompt -%}
136
+ {{- '<|im_start|>assistant
137
+ ' -}}
138
+ {%- endif -%}
config.json ADDED
@@ -0,0 +1,93 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen2VLForConditionalGeneration"
4
+ ],
5
+ "dtype": "float16",
6
+ "image_token_id": 151655,
7
+ "model_type": "qwen2_vl",
8
+ "text_config": {
9
+ "attention_dropout": 0.0,
10
+ "bos_token_id": 151643,
11
+ "dtype": "float16",
12
+ "eos_token_id": 151645,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 1536,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": 8960,
17
+ "layer_types": [
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention"
46
+ ],
47
+ "max_position_embeddings": 32768,
48
+ "max_window_layers": 28,
49
+ "model_type": "qwen2_vl_text",
50
+ "num_attention_heads": 12,
51
+ "num_hidden_layers": 28,
52
+ "num_key_value_heads": 2,
53
+ "pad_token_id": null,
54
+ "rms_norm_eps": 1e-06,
55
+ "rope_parameters": {
56
+ "mrope_section": [
57
+ 16,
58
+ 24,
59
+ 24
60
+ ],
61
+ "rope_theta": 1000000.0,
62
+ "rope_type": "default",
63
+ "type": "default"
64
+ },
65
+ "sliding_window": null,
66
+ "use_cache": true,
67
+ "use_sliding_window": false,
68
+ "vocab_size": 151936
69
+ },
70
+ "tie_word_embeddings": true,
71
+ "transformers_version": "5.3.0",
72
+ "video_token_id": 151656,
73
+ "vision_config": {
74
+ "depth": 32,
75
+ "dtype": "float16",
76
+ "embed_dim": 1280,
77
+ "hidden_act": "quick_gelu",
78
+ "hidden_size": 1536,
79
+ "in_channels": 3,
80
+ "in_chans": 3,
81
+ "initializer_range": 0.02,
82
+ "mlp_ratio": 4,
83
+ "model_type": "qwen2_vl",
84
+ "num_heads": 16,
85
+ "patch_size": 14,
86
+ "spatial_merge_size": 2,
87
+ "spatial_patch_size": 14,
88
+ "temporal_patch_size": 2
89
+ },
90
+ "vision_end_token_id": 151653,
91
+ "vision_start_token_id": 151652,
92
+ "vision_token_id": 151654
93
+ }
processor_config.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "image_processor": {
3
+ "data_format": "channels_first",
4
+ "do_convert_rgb": true,
5
+ "do_normalize": true,
6
+ "do_rescale": true,
7
+ "do_resize": true,
8
+ "image_mean": [
9
+ 0.48145466,
10
+ 0.4578275,
11
+ 0.40821073
12
+ ],
13
+ "image_processor_type": "Qwen2VLImageProcessorFast",
14
+ "image_std": [
15
+ 0.26862954,
16
+ 0.26130258,
17
+ 0.27577711
18
+ ],
19
+ "merge_size": 2,
20
+ "patch_size": 14,
21
+ "resample": 3,
22
+ "rescale_factor": 0.00392156862745098,
23
+ "size": {
24
+ "longest_edge": 301056,
25
+ "shortest_edge": 100352
26
+ },
27
+ "temporal_patch_size": 2
28
+ },
29
+ "processor_class": "Qwen2_5_VLProcessor",
30
+ "video_processor": {
31
+ "data_format": "channels_first",
32
+ "default_to_square": true,
33
+ "do_convert_rgb": true,
34
+ "do_normalize": true,
35
+ "do_rescale": true,
36
+ "do_resize": true,
37
+ "do_sample_frames": false,
38
+ "image_mean": [
39
+ 0.48145466,
40
+ 0.4578275,
41
+ 0.40821073
42
+ ],
43
+ "image_processor_type": "Qwen2VLImageProcessor",
44
+ "image_std": [
45
+ 0.26862954,
46
+ 0.26130258,
47
+ 0.27577711
48
+ ],
49
+ "max_frames": 768,
50
+ "merge_size": 2,
51
+ "min_frames": 4,
52
+ "patch_size": 14,
53
+ "resample": 3,
54
+ "rescale_factor": 0.00392156862745098,
55
+ "return_metadata": false,
56
+ "size": {
57
+ "longest_edge": 301056,
58
+ "shortest_edge": 100352
59
+ },
60
+ "temporal_patch_size": 2,
61
+ "video_processor_type": "Qwen2VLVideoProcessor"
62
+ }
63
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1a7aabe7a3b44651592a99607bb33066e6965260336be9218cccd53af161491d
3
+ size 11422060
tokenizer_config.json ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_commit_hash": null,
3
+ "add_prefix_space": false,
4
+ "backend": "tokenizers",
5
+ "bos_token": null,
6
+ "clean_up_tokenization_spaces": false,
7
+ "eos_token": "<|im_end|>",
8
+ "errors": "replace",
9
+ "is_local": true,
10
+ "max_length": null,
11
+ "max_pixels": 301056,
12
+ "min_pixels": 100352,
13
+ "model_max_length": 131072,
14
+ "pad_to_multiple_of": null,
15
+ "pad_token": "<|endoftext|>",
16
+ "pad_token_type_id": 0,
17
+ "padding_side": "right",
18
+ "processor_class": "Qwen2_5_VLProcessor",
19
+ "split_special_tokens": false,
20
+ "tokenizer_class": "Qwen2Tokenizer",
21
+ "unk_token": null
22
+ }
w8a8_smoothquant_meta.json ADDED
@@ -0,0 +1,2302 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "format_version": 1,
3
+ "artifact_type": "compressingvlm_w8a8_smoothquant_bundle",
4
+ "base_model": "/content/compressingvlm_nuextract2/artifacts/merged_full_checkpoint",
5
+ "dataset_id": "letxbe/BoundingDocs",
6
+ "smoothquant_alpha": 0.5,
7
+ "converted_layers": 327,
8
+ "skipped_layers": 0,
9
+ "quant_meta": {
10
+ "model.visual.blocks.0.attn.qkv": {
11
+ "in_features": 1280,
12
+ "out_features": 3840,
13
+ "act_scale": 0.011774783059367983,
14
+ "smoothquant_alpha": 0.5,
15
+ "out_dtype": "torch.float16"
16
+ },
17
+ "model.visual.blocks.0.attn.proj": {
18
+ "in_features": 1280,
19
+ "out_features": 1280,
20
+ "act_scale": 0.004796601186587116,
21
+ "smoothquant_alpha": 0.5,
22
+ "out_dtype": "torch.float16"
23
+ },
24
+ "model.visual.blocks.0.mlp.fc1": {
25
+ "in_features": 1280,
26
+ "out_features": 5120,
27
+ "act_scale": 0.020350467501662847,
28
+ "smoothquant_alpha": 0.5,
29
+ "out_dtype": "torch.float16"
30
+ },
31
+ "model.visual.blocks.0.mlp.fc2": {
32
+ "in_features": 5120,
33
+ "out_features": 1280,
34
+ "act_scale": 0.01206465593473179,
35
+ "smoothquant_alpha": 0.5,
36
+ "out_dtype": "torch.float16"
37
+ },
38
+ "model.visual.blocks.1.attn.qkv": {
39
+ "in_features": 1280,
40
+ "out_features": 3840,
41
+ "act_scale": 0.013214042806249904,
42
+ "smoothquant_alpha": 0.5,
43
+ "out_dtype": "torch.float16"
44
+ },
45
+ "model.visual.blocks.1.attn.proj": {
46
+ "in_features": 1280,
47
+ "out_features": 1280,
48
+ "act_scale": 0.00883366930203175,
49
+ "smoothquant_alpha": 0.5,
50
+ "out_dtype": "torch.float16"
51
+ },
52
+ "model.visual.blocks.1.mlp.fc1": {
53
+ "in_features": 1280,
54
+ "out_features": 5120,
55
+ "act_scale": 0.0195614765948198,
56
+ "smoothquant_alpha": 0.5,
57
+ "out_dtype": "torch.float16"
58
+ },
59
+ "model.visual.blocks.1.mlp.fc2": {
60
+ "in_features": 5120,
61
+ "out_features": 1280,
62
+ "act_scale": 0.006876293599136232,
63
+ "smoothquant_alpha": 0.5,
64
+ "out_dtype": "torch.float16"
65
+ },
66
+ "model.visual.blocks.2.attn.qkv": {
67
+ "in_features": 1280,
68
+ "out_features": 3840,
69
+ "act_scale": 0.016577895232072964,
70
+ "smoothquant_alpha": 0.5,
71
+ "out_dtype": "torch.float16"
72
+ },
73
+ "model.visual.blocks.2.attn.proj": {
74
+ "in_features": 1280,
75
+ "out_features": 1280,
76
+ "act_scale": 0.0066642273129440665,
77
+ "smoothquant_alpha": 0.5,
78
+ "out_dtype": "torch.float16"
79
+ },
80
+ "model.visual.blocks.2.mlp.fc1": {
81
+ "in_features": 1280,
82
+ "out_features": 5120,
83
+ "act_scale": 0.010180345670444759,
84
+ "smoothquant_alpha": 0.5,
85
+ "out_dtype": "torch.float16"
86
+ },
87
+ "model.visual.blocks.2.mlp.fc2": {
88
+ "in_features": 5120,
89
+ "out_features": 1280,
90
+ "act_scale": 0.007410509849157859,
91
+ "smoothquant_alpha": 0.5,
92
+ "out_dtype": "torch.float16"
93
+ },
94
+ "model.visual.blocks.3.attn.qkv": {
95
+ "in_features": 1280,
96
+ "out_features": 3840,
97
+ "act_scale": 0.020379057080726924,
98
+ "smoothquant_alpha": 0.5,
99
+ "out_dtype": "torch.float16"
100
+ },
101
+ "model.visual.blocks.3.attn.proj": {
102
+ "in_features": 1280,
103
+ "out_features": 1280,
104
+ "act_scale": 0.006451350028120626,
105
+ "smoothquant_alpha": 0.5,
106
+ "out_dtype": "torch.float16"
107
+ },
108
+ "model.visual.blocks.3.mlp.fc1": {
109
+ "in_features": 1280,
110
+ "out_features": 5120,
111
+ "act_scale": 0.010919001158766858,
112
+ "smoothquant_alpha": 0.5,
113
+ "out_dtype": "torch.float16"
114
+ },
115
+ "model.visual.blocks.3.mlp.fc2": {
116
+ "in_features": 5120,
117
+ "out_features": 1280,
118
+ "act_scale": 0.010403279244430422,
119
+ "smoothquant_alpha": 0.5,
120
+ "out_dtype": "torch.float16"
121
+ },
122
+ "model.visual.blocks.4.attn.qkv": {
123
+ "in_features": 1280,
124
+ "out_features": 3840,
125
+ "act_scale": 0.014471732725308635,
126
+ "smoothquant_alpha": 0.5,
127
+ "out_dtype": "torch.float16"
128
+ },
129
+ "model.visual.blocks.4.attn.proj": {
130
+ "in_features": 1280,
131
+ "out_features": 1280,
132
+ "act_scale": 0.007603761248701201,
133
+ "smoothquant_alpha": 0.5,
134
+ "out_dtype": "torch.float16"
135
+ },
136
+ "model.visual.blocks.4.mlp.fc1": {
137
+ "in_features": 1280,
138
+ "out_features": 5120,
139
+ "act_scale": 0.011414487530866007,
140
+ "smoothquant_alpha": 0.5,
141
+ "out_dtype": "torch.float16"
142
+ },
143
+ "model.visual.blocks.4.mlp.fc2": {
144
+ "in_features": 5120,
145
+ "out_features": 1280,
146
+ "act_scale": 0.008864887117400883,
147
+ "smoothquant_alpha": 0.5,
148
+ "out_dtype": "torch.float16"
149
+ },
150
+ "model.visual.blocks.5.attn.qkv": {
151
+ "in_features": 1280,
152
+ "out_features": 3840,
153
+ "act_scale": 0.009759143581540565,
154
+ "smoothquant_alpha": 0.5,
155
+ "out_dtype": "torch.float16"
156
+ },
157
+ "model.visual.blocks.5.attn.proj": {
158
+ "in_features": 1280,
159
+ "out_features": 1280,
160
+ "act_scale": 0.005876585723846916,
161
+ "smoothquant_alpha": 0.5,
162
+ "out_dtype": "torch.float16"
163
+ },
164
+ "model.visual.blocks.5.mlp.fc1": {
165
+ "in_features": 1280,
166
+ "out_features": 5120,
167
+ "act_scale": 0.0104806751716794,
168
+ "smoothquant_alpha": 0.5,
169
+ "out_dtype": "torch.float16"
170
+ },
171
+ "model.visual.blocks.5.mlp.fc2": {
172
+ "in_features": 5120,
173
+ "out_features": 1280,
174
+ "act_scale": 0.0067109634557108245,
175
+ "smoothquant_alpha": 0.5,
176
+ "out_dtype": "torch.float16"
177
+ },
178
+ "model.visual.blocks.6.attn.qkv": {
179
+ "in_features": 1280,
180
+ "out_features": 3840,
181
+ "act_scale": 0.011332700571675939,
182
+ "smoothquant_alpha": 0.5,
183
+ "out_dtype": "torch.float16"
184
+ },
185
+ "model.visual.blocks.6.attn.proj": {
186
+ "in_features": 1280,
187
+ "out_features": 1280,
188
+ "act_scale": 0.005560288279075322,
189
+ "smoothquant_alpha": 0.5,
190
+ "out_dtype": "torch.float16"
191
+ },
192
+ "model.visual.blocks.6.mlp.fc1": {
193
+ "in_features": 1280,
194
+ "out_features": 5120,
195
+ "act_scale": 0.014120476452384408,
196
+ "smoothquant_alpha": 0.5,
197
+ "out_dtype": "torch.float16"
198
+ },
199
+ "model.visual.blocks.6.mlp.fc2": {
200
+ "in_features": 5120,
201
+ "out_features": 1280,
202
+ "act_scale": 0.009994748070484071,
203
+ "smoothquant_alpha": 0.5,
204
+ "out_dtype": "torch.float16"
205
+ },
206
+ "model.visual.blocks.7.attn.qkv": {
207
+ "in_features": 1280,
208
+ "out_features": 3840,
209
+ "act_scale": 0.013232500534357988,
210
+ "smoothquant_alpha": 0.5,
211
+ "out_dtype": "torch.float16"
212
+ },
213
+ "model.visual.blocks.7.attn.proj": {
214
+ "in_features": 1280,
215
+ "out_features": 1280,
216
+ "act_scale": 0.005875116258155642,
217
+ "smoothquant_alpha": 0.5,
218
+ "out_dtype": "torch.float16"
219
+ },
220
+ "model.visual.blocks.7.mlp.fc1": {
221
+ "in_features": 1280,
222
+ "out_features": 5120,
223
+ "act_scale": 0.013313860405148484,
224
+ "smoothquant_alpha": 0.5,
225
+ "out_dtype": "torch.float16"
226
+ },
227
+ "model.visual.blocks.7.mlp.fc2": {
228
+ "in_features": 5120,
229
+ "out_features": 1280,
230
+ "act_scale": 0.008737835358446978,
231
+ "smoothquant_alpha": 0.5,
232
+ "out_dtype": "torch.float16"
233
+ },
234
+ "model.visual.blocks.8.attn.qkv": {
235
+ "in_features": 1280,
236
+ "out_features": 3840,
237
+ "act_scale": 0.021730391059334823,
238
+ "smoothquant_alpha": 0.5,
239
+ "out_dtype": "torch.float16"
240
+ },
241
+ "model.visual.blocks.8.attn.proj": {
242
+ "in_features": 1280,
243
+ "out_features": 1280,
244
+ "act_scale": 0.010192882357619878,
245
+ "smoothquant_alpha": 0.5,
246
+ "out_dtype": "torch.float16"
247
+ },
248
+ "model.visual.blocks.8.mlp.fc1": {
249
+ "in_features": 1280,
250
+ "out_features": 5120,
251
+ "act_scale": 0.016515345085324266,
252
+ "smoothquant_alpha": 0.5,
253
+ "out_dtype": "torch.float16"
254
+ },
255
+ "model.visual.blocks.8.mlp.fc2": {
256
+ "in_features": 5120,
257
+ "out_features": 1280,
258
+ "act_scale": 0.009404166476933037,
259
+ "smoothquant_alpha": 0.5,
260
+ "out_dtype": "torch.float16"
261
+ },
262
+ "model.visual.blocks.9.attn.qkv": {
263
+ "in_features": 1280,
264
+ "out_features": 3840,
265
+ "act_scale": 0.014161588638786256,
266
+ "smoothquant_alpha": 0.5,
267
+ "out_dtype": "torch.float16"
268
+ },
269
+ "model.visual.blocks.9.attn.proj": {
270
+ "in_features": 1280,
271
+ "out_features": 1280,
272
+ "act_scale": 0.005805544027193325,
273
+ "smoothquant_alpha": 0.5,
274
+ "out_dtype": "torch.float16"
275
+ },
276
+ "model.visual.blocks.9.mlp.fc1": {
277
+ "in_features": 1280,
278
+ "out_features": 5120,
279
+ "act_scale": 0.0160279217667467,
280
+ "smoothquant_alpha": 0.5,
281
+ "out_dtype": "torch.float16"
282
+ },
283
+ "model.visual.blocks.9.mlp.fc2": {
284
+ "in_features": 5120,
285
+ "out_features": 1280,
286
+ "act_scale": 0.012330186648631659,
287
+ "smoothquant_alpha": 0.5,
288
+ "out_dtype": "torch.float16"
289
+ },
290
+ "model.visual.blocks.10.attn.qkv": {
291
+ "in_features": 1280,
292
+ "out_features": 3840,
293
+ "act_scale": 0.01268474323543038,
294
+ "smoothquant_alpha": 0.5,
295
+ "out_dtype": "torch.float16"
296
+ },
297
+ "model.visual.blocks.10.attn.proj": {
298
+ "in_features": 1280,
299
+ "out_features": 1280,
300
+ "act_scale": 0.005399912830412857,
301
+ "smoothquant_alpha": 0.5,
302
+ "out_dtype": "torch.float16"
303
+ },
304
+ "model.visual.blocks.10.mlp.fc1": {
305
+ "in_features": 1280,
306
+ "out_features": 5120,
307
+ "act_scale": 0.015144372549582653,
308
+ "smoothquant_alpha": 0.5,
309
+ "out_dtype": "torch.float16"
310
+ },
311
+ "model.visual.blocks.10.mlp.fc2": {
312
+ "in_features": 5120,
313
+ "out_features": 1280,
314
+ "act_scale": 0.012361452335447777,
315
+ "smoothquant_alpha": 0.5,
316
+ "out_dtype": "torch.float16"
317
+ },
318
+ "model.visual.blocks.11.attn.qkv": {
319
+ "in_features": 1280,
320
+ "out_features": 3840,
321
+ "act_scale": 0.015269914949972799,
322
+ "smoothquant_alpha": 0.5,
323
+ "out_dtype": "torch.float16"
324
+ },
325
+ "model.visual.blocks.11.attn.proj": {
326
+ "in_features": 1280,
327
+ "out_features": 1280,
328
+ "act_scale": 0.006234276482439417,
329
+ "smoothquant_alpha": 0.5,
330
+ "out_dtype": "torch.float16"
331
+ },
332
+ "model.visual.blocks.11.mlp.fc1": {
333
+ "in_features": 1280,
334
+ "out_features": 5120,
335
+ "act_scale": 0.014402702098756324,
336
+ "smoothquant_alpha": 0.5,
337
+ "out_dtype": "torch.float16"
338
+ },
339
+ "model.visual.blocks.11.mlp.fc2": {
340
+ "in_features": 5120,
341
+ "out_features": 1280,
342
+ "act_scale": 0.01632130615354523,
343
+ "smoothquant_alpha": 0.5,
344
+ "out_dtype": "torch.float16"
345
+ },
346
+ "model.visual.blocks.12.attn.qkv": {
347
+ "in_features": 1280,
348
+ "out_features": 3840,
349
+ "act_scale": 0.014100967429754302,
350
+ "smoothquant_alpha": 0.5,
351
+ "out_dtype": "torch.float16"
352
+ },
353
+ "model.visual.blocks.12.attn.proj": {
354
+ "in_features": 1280,
355
+ "out_features": 1280,
356
+ "act_scale": 0.006146921416905921,
357
+ "smoothquant_alpha": 0.5,
358
+ "out_dtype": "torch.float16"
359
+ },
360
+ "model.visual.blocks.12.mlp.fc1": {
361
+ "in_features": 1280,
362
+ "out_features": 5120,
363
+ "act_scale": 0.016758920639518678,
364
+ "smoothquant_alpha": 0.5,
365
+ "out_dtype": "torch.float16"
366
+ },
367
+ "model.visual.blocks.12.mlp.fc2": {
368
+ "in_features": 5120,
369
+ "out_features": 1280,
370
+ "act_scale": 0.01607548533462164,
371
+ "smoothquant_alpha": 0.5,
372
+ "out_dtype": "torch.float16"
373
+ },
374
+ "model.visual.blocks.13.attn.qkv": {
375
+ "in_features": 1280,
376
+ "out_features": 3840,
377
+ "act_scale": 0.01547414486802469,
378
+ "smoothquant_alpha": 0.5,
379
+ "out_dtype": "torch.float16"
380
+ },
381
+ "model.visual.blocks.13.attn.proj": {
382
+ "in_features": 1280,
383
+ "out_features": 1280,
384
+ "act_scale": 0.0061614091002096345,
385
+ "smoothquant_alpha": 0.5,
386
+ "out_dtype": "torch.float16"
387
+ },
388
+ "model.visual.blocks.13.mlp.fc1": {
389
+ "in_features": 1280,
390
+ "out_features": 5120,
391
+ "act_scale": 0.01329935817267951,
392
+ "smoothquant_alpha": 0.5,
393
+ "out_dtype": "torch.float16"
394
+ },
395
+ "model.visual.blocks.13.mlp.fc2": {
396
+ "in_features": 5120,
397
+ "out_features": 1280,
398
+ "act_scale": 0.0183927956528551,
399
+ "smoothquant_alpha": 0.5,
400
+ "out_dtype": "torch.float16"
401
+ },
402
+ "model.visual.blocks.14.attn.qkv": {
403
+ "in_features": 1280,
404
+ "out_features": 3840,
405
+ "act_scale": 0.016157283557681586,
406
+ "smoothquant_alpha": 0.5,
407
+ "out_dtype": "torch.float16"
408
+ },
409
+ "model.visual.blocks.14.attn.proj": {
410
+ "in_features": 1280,
411
+ "out_features": 1280,
412
+ "act_scale": 0.0056524656889006845,
413
+ "smoothquant_alpha": 0.5,
414
+ "out_dtype": "torch.float16"
415
+ },
416
+ "model.visual.blocks.14.mlp.fc1": {
417
+ "in_features": 1280,
418
+ "out_features": 5120,
419
+ "act_scale": 0.014731704719423309,
420
+ "smoothquant_alpha": 0.5,
421
+ "out_dtype": "torch.float16"
422
+ },
423
+ "model.visual.blocks.14.mlp.fc2": {
424
+ "in_features": 5120,
425
+ "out_features": 1280,
426
+ "act_scale": 0.0164302334072083,
427
+ "smoothquant_alpha": 0.5,
428
+ "out_dtype": "torch.float16"
429
+ },
430
+ "model.visual.blocks.15.attn.qkv": {
431
+ "in_features": 1280,
432
+ "out_features": 3840,
433
+ "act_scale": 0.015307413311455194,
434
+ "smoothquant_alpha": 0.5,
435
+ "out_dtype": "torch.float16"
436
+ },
437
+ "model.visual.blocks.15.attn.proj": {
438
+ "in_features": 1280,
439
+ "out_features": 1280,
440
+ "act_scale": 0.006376280559329536,
441
+ "smoothquant_alpha": 0.5,
442
+ "out_dtype": "torch.float16"
443
+ },
444
+ "model.visual.blocks.15.mlp.fc1": {
445
+ "in_features": 1280,
446
+ "out_features": 5120,
447
+ "act_scale": 0.015562512743191457,
448
+ "smoothquant_alpha": 0.5,
449
+ "out_dtype": "torch.float16"
450
+ },
451
+ "model.visual.blocks.15.mlp.fc2": {
452
+ "in_features": 5120,
453
+ "out_features": 1280,
454
+ "act_scale": 0.025342841786662424,
455
+ "smoothquant_alpha": 0.5,
456
+ "out_dtype": "torch.float16"
457
+ },
458
+ "model.visual.blocks.16.attn.qkv": {
459
+ "in_features": 1280,
460
+ "out_features": 3840,
461
+ "act_scale": 0.015025780895563562,
462
+ "smoothquant_alpha": 0.5,
463
+ "out_dtype": "torch.float16"
464
+ },
465
+ "model.visual.blocks.16.attn.proj": {
466
+ "in_features": 1280,
467
+ "out_features": 1280,
468
+ "act_scale": 0.005419237407173697,
469
+ "smoothquant_alpha": 0.5,
470
+ "out_dtype": "torch.float16"
471
+ },
472
+ "model.visual.blocks.16.mlp.fc1": {
473
+ "in_features": 1280,
474
+ "out_features": 5120,
475
+ "act_scale": 0.016547708060797743,
476
+ "smoothquant_alpha": 0.5,
477
+ "out_dtype": "torch.float16"
478
+ },
479
+ "model.visual.blocks.16.mlp.fc2": {
480
+ "in_features": 5120,
481
+ "out_features": 1280,
482
+ "act_scale": 0.018387858323224888,
483
+ "smoothquant_alpha": 0.5,
484
+ "out_dtype": "torch.float16"
485
+ },
486
+ "model.visual.blocks.17.attn.qkv": {
487
+ "in_features": 1280,
488
+ "out_features": 3840,
489
+ "act_scale": 0.01265057522480882,
490
+ "smoothquant_alpha": 0.5,
491
+ "out_dtype": "torch.float16"
492
+ },
493
+ "model.visual.blocks.17.attn.proj": {
494
+ "in_features": 1280,
495
+ "out_features": 1280,
496
+ "act_scale": 0.006702738484059732,
497
+ "smoothquant_alpha": 0.5,
498
+ "out_dtype": "torch.float16"
499
+ },
500
+ "model.visual.blocks.17.mlp.fc1": {
501
+ "in_features": 1280,
502
+ "out_features": 5120,
503
+ "act_scale": 0.02144298027819536,
504
+ "smoothquant_alpha": 0.5,
505
+ "out_dtype": "torch.float16"
506
+ },
507
+ "model.visual.blocks.17.mlp.fc2": {
508
+ "in_features": 5120,
509
+ "out_features": 1280,
510
+ "act_scale": 0.02630904340368556,
511
+ "smoothquant_alpha": 0.5,
512
+ "out_dtype": "torch.float16"
513
+ },
514
+ "model.visual.blocks.18.attn.qkv": {
515
+ "in_features": 1280,
516
+ "out_features": 3840,
517
+ "act_scale": 0.014642170095068264,
518
+ "smoothquant_alpha": 0.5,
519
+ "out_dtype": "torch.float16"
520
+ },
521
+ "model.visual.blocks.18.attn.proj": {
522
+ "in_features": 1280,
523
+ "out_features": 1280,
524
+ "act_scale": 0.00903098601994552,
525
+ "smoothquant_alpha": 0.5,
526
+ "out_dtype": "torch.float16"
527
+ },
528
+ "model.visual.blocks.18.mlp.fc1": {
529
+ "in_features": 1280,
530
+ "out_features": 5120,
531
+ "act_scale": 0.019541541422445942,
532
+ "smoothquant_alpha": 0.5,
533
+ "out_dtype": "torch.float16"
534
+ },
535
+ "model.visual.blocks.18.mlp.fc2": {
536
+ "in_features": 5120,
537
+ "out_features": 1280,
538
+ "act_scale": 0.04239599348053219,
539
+ "smoothquant_alpha": 0.5,
540
+ "out_dtype": "torch.float16"
541
+ },
542
+ "model.visual.blocks.19.attn.qkv": {
543
+ "in_features": 1280,
544
+ "out_features": 3840,
545
+ "act_scale": 0.014888684580645224,
546
+ "smoothquant_alpha": 0.5,
547
+ "out_dtype": "torch.float16"
548
+ },
549
+ "model.visual.blocks.19.attn.proj": {
550
+ "in_features": 1280,
551
+ "out_features": 1280,
552
+ "act_scale": 0.006763935089111328,
553
+ "smoothquant_alpha": 0.5,
554
+ "out_dtype": "torch.float16"
555
+ },
556
+ "model.visual.blocks.19.mlp.fc1": {
557
+ "in_features": 1280,
558
+ "out_features": 5120,
559
+ "act_scale": 0.01819218237569013,
560
+ "smoothquant_alpha": 0.5,
561
+ "out_dtype": "torch.float16"
562
+ },
563
+ "model.visual.blocks.19.mlp.fc2": {
564
+ "in_features": 5120,
565
+ "out_features": 1280,
566
+ "act_scale": 0.01527449746770183,
567
+ "smoothquant_alpha": 0.5,
568
+ "out_dtype": "torch.float16"
569
+ },
570
+ "model.visual.blocks.20.attn.qkv": {
571
+ "in_features": 1280,
572
+ "out_features": 3840,
573
+ "act_scale": 0.014723354437219815,
574
+ "smoothquant_alpha": 0.5,
575
+ "out_dtype": "torch.float16"
576
+ },
577
+ "model.visual.blocks.20.attn.proj": {
578
+ "in_features": 1280,
579
+ "out_features": 1280,
580
+ "act_scale": 0.006135502199488362,
581
+ "smoothquant_alpha": 0.5,
582
+ "out_dtype": "torch.float16"
583
+ },
584
+ "model.visual.blocks.20.mlp.fc1": {
585
+ "in_features": 1280,
586
+ "out_features": 5120,
587
+ "act_scale": 0.019845838621845396,
588
+ "smoothquant_alpha": 0.5,
589
+ "out_dtype": "torch.float16"
590
+ },
591
+ "model.visual.blocks.20.mlp.fc2": {
592
+ "in_features": 5120,
593
+ "out_features": 1280,
594
+ "act_scale": 0.013107329841673844,
595
+ "smoothquant_alpha": 0.5,
596
+ "out_dtype": "torch.float16"
597
+ },
598
+ "model.visual.blocks.21.attn.qkv": {
599
+ "in_features": 1280,
600
+ "out_features": 3840,
601
+ "act_scale": 0.013580497794263945,
602
+ "smoothquant_alpha": 0.5,
603
+ "out_dtype": "torch.float16"
604
+ },
605
+ "model.visual.blocks.21.attn.proj": {
606
+ "in_features": 1280,
607
+ "out_features": 1280,
608
+ "act_scale": 0.012507303493229423,
609
+ "smoothquant_alpha": 0.5,
610
+ "out_dtype": "torch.float16"
611
+ },
612
+ "model.visual.blocks.21.mlp.fc1": {
613
+ "in_features": 1280,
614
+ "out_features": 5120,
615
+ "act_scale": 0.018828945835744303,
616
+ "smoothquant_alpha": 0.5,
617
+ "out_dtype": "torch.float16"
618
+ },
619
+ "model.visual.blocks.21.mlp.fc2": {
620
+ "in_features": 5120,
621
+ "out_features": 1280,
622
+ "act_scale": 0.019425249475193775,
623
+ "smoothquant_alpha": 0.5,
624
+ "out_dtype": "torch.float16"
625
+ },
626
+ "model.visual.blocks.22.attn.qkv": {
627
+ "in_features": 1280,
628
+ "out_features": 3840,
629
+ "act_scale": 0.014368930200892171,
630
+ "smoothquant_alpha": 0.5,
631
+ "out_dtype": "torch.float16"
632
+ },
633
+ "model.visual.blocks.22.attn.proj": {
634
+ "in_features": 1280,
635
+ "out_features": 1280,
636
+ "act_scale": 0.008607041178725836,
637
+ "smoothquant_alpha": 0.5,
638
+ "out_dtype": "torch.float16"
639
+ },
640
+ "model.visual.blocks.22.mlp.fc1": {
641
+ "in_features": 1280,
642
+ "out_features": 5120,
643
+ "act_scale": 0.020806425199733945,
644
+ "smoothquant_alpha": 0.5,
645
+ "out_dtype": "torch.float16"
646
+ },
647
+ "model.visual.blocks.22.mlp.fc2": {
648
+ "in_features": 5120,
649
+ "out_features": 1280,
650
+ "act_scale": 0.0203628277215432,
651
+ "smoothquant_alpha": 0.5,
652
+ "out_dtype": "torch.float16"
653
+ },
654
+ "model.visual.blocks.23.attn.qkv": {
655
+ "in_features": 1280,
656
+ "out_features": 3840,
657
+ "act_scale": 0.015795058152807042,
658
+ "smoothquant_alpha": 0.5,
659
+ "out_dtype": "torch.float16"
660
+ },
661
+ "model.visual.blocks.23.attn.proj": {
662
+ "in_features": 1280,
663
+ "out_features": 1280,
664
+ "act_scale": 0.006380434111347349,
665
+ "smoothquant_alpha": 0.5,
666
+ "out_dtype": "torch.float16"
667
+ },
668
+ "model.visual.blocks.23.mlp.fc1": {
669
+ "in_features": 1280,
670
+ "out_features": 5120,
671
+ "act_scale": 0.023131884927824725,
672
+ "smoothquant_alpha": 0.5,
673
+ "out_dtype": "torch.float16"
674
+ },
675
+ "model.visual.blocks.23.mlp.fc2": {
676
+ "in_features": 5120,
677
+ "out_features": 1280,
678
+ "act_scale": 0.025056778915285124,
679
+ "smoothquant_alpha": 0.5,
680
+ "out_dtype": "torch.float16"
681
+ },
682
+ "model.visual.blocks.24.attn.qkv": {
683
+ "in_features": 1280,
684
+ "out_features": 3840,
685
+ "act_scale": 0.016849420202060008,
686
+ "smoothquant_alpha": 0.5,
687
+ "out_dtype": "torch.float16"
688
+ },
689
+ "model.visual.blocks.24.attn.proj": {
690
+ "in_features": 1280,
691
+ "out_features": 1280,
692
+ "act_scale": 0.007182589196783351,
693
+ "smoothquant_alpha": 0.5,
694
+ "out_dtype": "torch.float16"
695
+ },
696
+ "model.visual.blocks.24.mlp.fc1": {
697
+ "in_features": 1280,
698
+ "out_features": 5120,
699
+ "act_scale": 0.02979039394949365,
700
+ "smoothquant_alpha": 0.5,
701
+ "out_dtype": "torch.float16"
702
+ },
703
+ "model.visual.blocks.24.mlp.fc2": {
704
+ "in_features": 5120,
705
+ "out_features": 1280,
706
+ "act_scale": 0.02414914003507359,
707
+ "smoothquant_alpha": 0.5,
708
+ "out_dtype": "torch.float16"
709
+ },
710
+ "model.visual.blocks.25.attn.qkv": {
711
+ "in_features": 1280,
712
+ "out_features": 3840,
713
+ "act_scale": 0.015637225053441808,
714
+ "smoothquant_alpha": 0.5,
715
+ "out_dtype": "torch.float16"
716
+ },
717
+ "model.visual.blocks.25.attn.proj": {
718
+ "in_features": 1280,
719
+ "out_features": 1280,
720
+ "act_scale": 0.00804293437266913,
721
+ "smoothquant_alpha": 0.5,
722
+ "out_dtype": "torch.float16"
723
+ },
724
+ "model.visual.blocks.25.mlp.fc1": {
725
+ "in_features": 1280,
726
+ "out_features": 5120,
727
+ "act_scale": 0.0279038853532686,
728
+ "smoothquant_alpha": 0.5,
729
+ "out_dtype": "torch.float16"
730
+ },
731
+ "model.visual.blocks.25.mlp.fc2": {
732
+ "in_features": 5120,
733
+ "out_features": 1280,
734
+ "act_scale": 0.019783076338880642,
735
+ "smoothquant_alpha": 0.5,
736
+ "out_dtype": "torch.float16"
737
+ },
738
+ "model.visual.blocks.26.attn.qkv": {
739
+ "in_features": 1280,
740
+ "out_features": 3840,
741
+ "act_scale": 0.015581223908371813,
742
+ "smoothquant_alpha": 0.5,
743
+ "out_dtype": "torch.float16"
744
+ },
745
+ "model.visual.blocks.26.attn.proj": {
746
+ "in_features": 1280,
747
+ "out_features": 1280,
748
+ "act_scale": 0.008657125037486158,
749
+ "smoothquant_alpha": 0.5,
750
+ "out_dtype": "torch.float16"
751
+ },
752
+ "model.visual.blocks.26.mlp.fc1": {
753
+ "in_features": 1280,
754
+ "out_features": 5120,
755
+ "act_scale": 0.0347223582230215,
756
+ "smoothquant_alpha": 0.5,
757
+ "out_dtype": "torch.float16"
758
+ },
759
+ "model.visual.blocks.26.mlp.fc2": {
760
+ "in_features": 5120,
761
+ "out_features": 1280,
762
+ "act_scale": 0.019559443466306673,
763
+ "smoothquant_alpha": 0.5,
764
+ "out_dtype": "torch.float16"
765
+ },
766
+ "model.visual.blocks.27.attn.qkv": {
767
+ "in_features": 1280,
768
+ "out_features": 3840,
769
+ "act_scale": 0.015775372662882165,
770
+ "smoothquant_alpha": 0.5,
771
+ "out_dtype": "torch.float16"
772
+ },
773
+ "model.visual.blocks.27.attn.proj": {
774
+ "in_features": 1280,
775
+ "out_features": 1280,
776
+ "act_scale": 0.006993105561714473,
777
+ "smoothquant_alpha": 0.5,
778
+ "out_dtype": "torch.float16"
779
+ },
780
+ "model.visual.blocks.27.mlp.fc1": {
781
+ "in_features": 1280,
782
+ "out_features": 5120,
783
+ "act_scale": 0.0269776197869008,
784
+ "smoothquant_alpha": 0.5,
785
+ "out_dtype": "torch.float16"
786
+ },
787
+ "model.visual.blocks.27.mlp.fc2": {
788
+ "in_features": 5120,
789
+ "out_features": 1280,
790
+ "act_scale": 0.024150317109475926,
791
+ "smoothquant_alpha": 0.5,
792
+ "out_dtype": "torch.float16"
793
+ },
794
+ "model.visual.blocks.28.attn.qkv": {
795
+ "in_features": 1280,
796
+ "out_features": 3840,
797
+ "act_scale": 0.01766562649584192,
798
+ "smoothquant_alpha": 0.5,
799
+ "out_dtype": "torch.float16"
800
+ },
801
+ "model.visual.blocks.28.attn.proj": {
802
+ "in_features": 1280,
803
+ "out_features": 1280,
804
+ "act_scale": 0.008466911128186804,
805
+ "smoothquant_alpha": 0.5,
806
+ "out_dtype": "torch.float16"
807
+ },
808
+ "model.visual.blocks.28.mlp.fc1": {
809
+ "in_features": 1280,
810
+ "out_features": 5120,
811
+ "act_scale": 0.03087514404236801,
812
+ "smoothquant_alpha": 0.5,
813
+ "out_dtype": "torch.float16"
814
+ },
815
+ "model.visual.blocks.28.mlp.fc2": {
816
+ "in_features": 5120,
817
+ "out_features": 1280,
818
+ "act_scale": 0.026798460427231677,
819
+ "smoothquant_alpha": 0.5,
820
+ "out_dtype": "torch.float16"
821
+ },
822
+ "model.visual.blocks.29.attn.qkv": {
823
+ "in_features": 1280,
824
+ "out_features": 3840,
825
+ "act_scale": 0.017204213330126183,
826
+ "smoothquant_alpha": 0.5,
827
+ "out_dtype": "torch.float16"
828
+ },
829
+ "model.visual.blocks.29.attn.proj": {
830
+ "in_features": 1280,
831
+ "out_features": 1280,
832
+ "act_scale": 0.01030280834107887,
833
+ "smoothquant_alpha": 0.5,
834
+ "out_dtype": "torch.float16"
835
+ },
836
+ "model.visual.blocks.29.mlp.fc1": {
837
+ "in_features": 1280,
838
+ "out_features": 5120,
839
+ "act_scale": 0.03364264495729461,
840
+ "smoothquant_alpha": 0.5,
841
+ "out_dtype": "torch.float16"
842
+ },
843
+ "model.visual.blocks.29.mlp.fc2": {
844
+ "in_features": 5120,
845
+ "out_features": 1280,
846
+ "act_scale": 0.039391221023920016,
847
+ "smoothquant_alpha": 0.5,
848
+ "out_dtype": "torch.float16"
849
+ },
850
+ "model.visual.blocks.30.attn.qkv": {
851
+ "in_features": 1280,
852
+ "out_features": 3840,
853
+ "act_scale": 0.01684388964194951,
854
+ "smoothquant_alpha": 0.5,
855
+ "out_dtype": "torch.float16"
856
+ },
857
+ "model.visual.blocks.30.attn.proj": {
858
+ "in_features": 1280,
859
+ "out_features": 1280,
860
+ "act_scale": 0.013944210968618318,
861
+ "smoothquant_alpha": 0.5,
862
+ "out_dtype": "torch.float16"
863
+ },
864
+ "model.visual.blocks.30.mlp.fc1": {
865
+ "in_features": 1280,
866
+ "out_features": 5120,
867
+ "act_scale": 0.026909640454870508,
868
+ "smoothquant_alpha": 0.5,
869
+ "out_dtype": "torch.float16"
870
+ },
871
+ "model.visual.blocks.30.mlp.fc2": {
872
+ "in_features": 5120,
873
+ "out_features": 1280,
874
+ "act_scale": 0.032339970896563194,
875
+ "smoothquant_alpha": 0.5,
876
+ "out_dtype": "torch.float16"
877
+ },
878
+ "model.visual.blocks.31.attn.qkv": {
879
+ "in_features": 1280,
880
+ "out_features": 3840,
881
+ "act_scale": 0.013929445912518839,
882
+ "smoothquant_alpha": 0.5,
883
+ "out_dtype": "torch.float16"
884
+ },
885
+ "model.visual.blocks.31.attn.proj": {
886
+ "in_features": 1280,
887
+ "out_features": 1280,
888
+ "act_scale": 0.01802794576629879,
889
+ "smoothquant_alpha": 0.5,
890
+ "out_dtype": "torch.float16"
891
+ },
892
+ "model.visual.blocks.31.mlp.fc1": {
893
+ "in_features": 1280,
894
+ "out_features": 5120,
895
+ "act_scale": 0.01583776699276421,
896
+ "smoothquant_alpha": 0.5,
897
+ "out_dtype": "torch.float16"
898
+ },
899
+ "model.visual.blocks.31.mlp.fc2": {
900
+ "in_features": 5120,
901
+ "out_features": 1280,
902
+ "act_scale": 0.16578437024214138,
903
+ "smoothquant_alpha": 0.5,
904
+ "out_dtype": "torch.float16"
905
+ },
906
+ "model.visual.merger.mlp.0": {
907
+ "in_features": 5120,
908
+ "out_features": 5120,
909
+ "act_scale": 0.009428932910829078,
910
+ "smoothquant_alpha": 0.5,
911
+ "out_dtype": "torch.float16"
912
+ },
913
+ "model.visual.merger.mlp.2": {
914
+ "in_features": 5120,
915
+ "out_features": 1536,
916
+ "act_scale": 0.015239587918979915,
917
+ "smoothquant_alpha": 0.5,
918
+ "out_dtype": "torch.float16"
919
+ },
920
+ "model.language_model.layers.0.self_attn.q_proj": {
921
+ "in_features": 1536,
922
+ "out_features": 1536,
923
+ "act_scale": 0.038241622954841674,
924
+ "smoothquant_alpha": 0.5,
925
+ "out_dtype": "torch.float16"
926
+ },
927
+ "model.language_model.layers.0.self_attn.k_proj": {
928
+ "in_features": 1536,
929
+ "out_features": 256,
930
+ "act_scale": 0.019962245085107997,
931
+ "smoothquant_alpha": 0.5,
932
+ "out_dtype": "torch.float16"
933
+ },
934
+ "model.language_model.layers.0.self_attn.v_proj": {
935
+ "in_features": 1536,
936
+ "out_features": 256,
937
+ "act_scale": 0.008818717453423448,
938
+ "smoothquant_alpha": 0.5,
939
+ "out_dtype": "torch.float16"
940
+ },
941
+ "model.language_model.layers.0.self_attn.o_proj": {
942
+ "in_features": 1536,
943
+ "out_features": 1536,
944
+ "act_scale": 0.0067325891472223235,
945
+ "smoothquant_alpha": 0.5,
946
+ "out_dtype": "torch.float16"
947
+ },
948
+ "model.language_model.layers.0.mlp.gate_proj": {
949
+ "in_features": 1536,
950
+ "out_features": 8960,
951
+ "act_scale": 0.011143252605528344,
952
+ "smoothquant_alpha": 0.5,
953
+ "out_dtype": "torch.float16"
954
+ },
955
+ "model.language_model.layers.0.mlp.up_proj": {
956
+ "in_features": 1536,
957
+ "out_features": 8960,
958
+ "act_scale": 0.008632640200337088,
959
+ "smoothquant_alpha": 0.5,
960
+ "out_dtype": "torch.float16"
961
+ },
962
+ "model.language_model.layers.0.mlp.down_proj": {
963
+ "in_features": 8960,
964
+ "out_features": 1536,
965
+ "act_scale": 0.00833494362868662,
966
+ "smoothquant_alpha": 0.5,
967
+ "out_dtype": "torch.float16"
968
+ },
969
+ "model.language_model.layers.1.self_attn.q_proj": {
970
+ "in_features": 1536,
971
+ "out_features": 1536,
972
+ "act_scale": 0.015791344830370323,
973
+ "smoothquant_alpha": 0.5,
974
+ "out_dtype": "torch.float16"
975
+ },
976
+ "model.language_model.layers.1.self_attn.k_proj": {
977
+ "in_features": 1536,
978
+ "out_features": 256,
979
+ "act_scale": 0.012920356172276295,
980
+ "smoothquant_alpha": 0.5,
981
+ "out_dtype": "torch.float16"
982
+ },
983
+ "model.language_model.layers.1.self_attn.v_proj": {
984
+ "in_features": 1536,
985
+ "out_features": 256,
986
+ "act_scale": 0.007185041904449463,
987
+ "smoothquant_alpha": 0.5,
988
+ "out_dtype": "torch.float16"
989
+ },
990
+ "model.language_model.layers.1.self_attn.o_proj": {
991
+ "in_features": 1536,
992
+ "out_features": 1536,
993
+ "act_scale": 0.006423686902354083,
994
+ "smoothquant_alpha": 0.5,
995
+ "out_dtype": "torch.float16"
996
+ },
997
+ "model.language_model.layers.1.mlp.gate_proj": {
998
+ "in_features": 1536,
999
+ "out_features": 8960,
1000
+ "act_scale": 0.014584682119174266,
1001
+ "smoothquant_alpha": 0.5,
1002
+ "out_dtype": "torch.float16"
1003
+ },
1004
+ "model.language_model.layers.1.mlp.up_proj": {
1005
+ "in_features": 1536,
1006
+ "out_features": 8960,
1007
+ "act_scale": 0.021518359972736028,
1008
+ "smoothquant_alpha": 0.5,
1009
+ "out_dtype": "torch.float16"
1010
+ },
1011
+ "model.language_model.layers.1.mlp.down_proj": {
1012
+ "in_features": 8960,
1013
+ "out_features": 1536,
1014
+ "act_scale": 0.2898354718065637,
1015
+ "smoothquant_alpha": 0.5,
1016
+ "out_dtype": "torch.float16"
1017
+ },
1018
+ "model.language_model.layers.2.self_attn.q_proj": {
1019
+ "in_features": 1536,
1020
+ "out_features": 1536,
1021
+ "act_scale": 0.019355892196414978,
1022
+ "smoothquant_alpha": 0.5,
1023
+ "out_dtype": "torch.float16"
1024
+ },
1025
+ "model.language_model.layers.2.self_attn.k_proj": {
1026
+ "in_features": 1536,
1027
+ "out_features": 256,
1028
+ "act_scale": 0.01587051293981357,
1029
+ "smoothquant_alpha": 0.5,
1030
+ "out_dtype": "torch.float16"
1031
+ },
1032
+ "model.language_model.layers.2.self_attn.v_proj": {
1033
+ "in_features": 1536,
1034
+ "out_features": 256,
1035
+ "act_scale": 0.006699156104110357,
1036
+ "smoothquant_alpha": 0.5,
1037
+ "out_dtype": "torch.float16"
1038
+ },
1039
+ "model.language_model.layers.2.self_attn.o_proj": {
1040
+ "in_features": 1536,
1041
+ "out_features": 1536,
1042
+ "act_scale": 0.006997298067948949,
1043
+ "smoothquant_alpha": 0.5,
1044
+ "out_dtype": "torch.float16"
1045
+ },
1046
+ "model.language_model.layers.2.mlp.gate_proj": {
1047
+ "in_features": 1536,
1048
+ "out_features": 8960,
1049
+ "act_scale": 0.014152612273148664,
1050
+ "smoothquant_alpha": 0.5,
1051
+ "out_dtype": "torch.float16"
1052
+ },
1053
+ "model.language_model.layers.2.mlp.up_proj": {
1054
+ "in_features": 1536,
1055
+ "out_features": 8960,
1056
+ "act_scale": 0.015616326820193312,
1057
+ "smoothquant_alpha": 0.5,
1058
+ "out_dtype": "torch.float16"
1059
+ },
1060
+ "model.language_model.layers.2.mlp.down_proj": {
1061
+ "in_features": 8960,
1062
+ "out_features": 1536,
1063
+ "act_scale": 0.025589364720141793,
1064
+ "smoothquant_alpha": 0.5,
1065
+ "out_dtype": "torch.float16"
1066
+ },
1067
+ "model.language_model.layers.3.self_attn.q_proj": {
1068
+ "in_features": 1536,
1069
+ "out_features": 1536,
1070
+ "act_scale": 0.016779912738349494,
1071
+ "smoothquant_alpha": 0.5,
1072
+ "out_dtype": "torch.float16"
1073
+ },
1074
+ "model.language_model.layers.3.self_attn.k_proj": {
1075
+ "in_features": 1536,
1076
+ "out_features": 256,
1077
+ "act_scale": 0.014279944690193717,
1078
+ "smoothquant_alpha": 0.5,
1079
+ "out_dtype": "torch.float16"
1080
+ },
1081
+ "model.language_model.layers.3.self_attn.v_proj": {
1082
+ "in_features": 1536,
1083
+ "out_features": 256,
1084
+ "act_scale": 0.007599234111665741,
1085
+ "smoothquant_alpha": 0.5,
1086
+ "out_dtype": "torch.float16"
1087
+ },
1088
+ "model.language_model.layers.3.self_attn.o_proj": {
1089
+ "in_features": 1536,
1090
+ "out_features": 1536,
1091
+ "act_scale": 0.008523108452323853,
1092
+ "smoothquant_alpha": 0.5,
1093
+ "out_dtype": "torch.float16"
1094
+ },
1095
+ "model.language_model.layers.3.mlp.gate_proj": {
1096
+ "in_features": 1536,
1097
+ "out_features": 8960,
1098
+ "act_scale": 0.015464061827171506,
1099
+ "smoothquant_alpha": 0.5,
1100
+ "out_dtype": "torch.float16"
1101
+ },
1102
+ "model.language_model.layers.3.mlp.up_proj": {
1103
+ "in_features": 1536,
1104
+ "out_features": 8960,
1105
+ "act_scale": 0.021024826004749208,
1106
+ "smoothquant_alpha": 0.5,
1107
+ "out_dtype": "torch.float16"
1108
+ },
1109
+ "model.language_model.layers.3.mlp.down_proj": {
1110
+ "in_features": 8960,
1111
+ "out_features": 1536,
1112
+ "act_scale": 0.057624861950010764,
1113
+ "smoothquant_alpha": 0.5,
1114
+ "out_dtype": "torch.float16"
1115
+ },
1116
+ "model.language_model.layers.4.self_attn.q_proj": {
1117
+ "in_features": 1536,
1118
+ "out_features": 1536,
1119
+ "act_scale": 0.016623001399002676,
1120
+ "smoothquant_alpha": 0.5,
1121
+ "out_dtype": "torch.float16"
1122
+ },
1123
+ "model.language_model.layers.4.self_attn.k_proj": {
1124
+ "in_features": 1536,
1125
+ "out_features": 256,
1126
+ "act_scale": 0.013275729389641229,
1127
+ "smoothquant_alpha": 0.5,
1128
+ "out_dtype": "torch.float16"
1129
+ },
1130
+ "model.language_model.layers.4.self_attn.v_proj": {
1131
+ "in_features": 1536,
1132
+ "out_features": 256,
1133
+ "act_scale": 0.008648515686275452,
1134
+ "smoothquant_alpha": 0.5,
1135
+ "out_dtype": "torch.float16"
1136
+ },
1137
+ "model.language_model.layers.4.self_attn.o_proj": {
1138
+ "in_features": 1536,
1139
+ "out_features": 1536,
1140
+ "act_scale": 0.007065747197218767,
1141
+ "smoothquant_alpha": 0.5,
1142
+ "out_dtype": "torch.float16"
1143
+ },
1144
+ "model.language_model.layers.4.mlp.gate_proj": {
1145
+ "in_features": 1536,
1146
+ "out_features": 8960,
1147
+ "act_scale": 0.015994242795809046,
1148
+ "smoothquant_alpha": 0.5,
1149
+ "out_dtype": "torch.float16"
1150
+ },
1151
+ "model.language_model.layers.4.mlp.up_proj": {
1152
+ "in_features": 1536,
1153
+ "out_features": 8960,
1154
+ "act_scale": 0.016403669447410764,
1155
+ "smoothquant_alpha": 0.5,
1156
+ "out_dtype": "torch.float16"
1157
+ },
1158
+ "model.language_model.layers.4.mlp.down_proj": {
1159
+ "in_features": 8960,
1160
+ "out_features": 1536,
1161
+ "act_scale": 0.01744353865075299,
1162
+ "smoothquant_alpha": 0.5,
1163
+ "out_dtype": "torch.float16"
1164
+ },
1165
+ "model.language_model.layers.5.self_attn.q_proj": {
1166
+ "in_features": 1536,
1167
+ "out_features": 1536,
1168
+ "act_scale": 0.018114063683457263,
1169
+ "smoothquant_alpha": 0.5,
1170
+ "out_dtype": "torch.float16"
1171
+ },
1172
+ "model.language_model.layers.5.self_attn.k_proj": {
1173
+ "in_features": 1536,
1174
+ "out_features": 256,
1175
+ "act_scale": 0.013239755405215766,
1176
+ "smoothquant_alpha": 0.5,
1177
+ "out_dtype": "torch.float16"
1178
+ },
1179
+ "model.language_model.layers.5.self_attn.v_proj": {
1180
+ "in_features": 1536,
1181
+ "out_features": 256,
1182
+ "act_scale": 0.009236081378666435,
1183
+ "smoothquant_alpha": 0.5,
1184
+ "out_dtype": "torch.float16"
1185
+ },
1186
+ "model.language_model.layers.5.self_attn.o_proj": {
1187
+ "in_features": 1536,
1188
+ "out_features": 1536,
1189
+ "act_scale": 0.0064572513572813016,
1190
+ "smoothquant_alpha": 0.5,
1191
+ "out_dtype": "torch.float16"
1192
+ },
1193
+ "model.language_model.layers.5.mlp.gate_proj": {
1194
+ "in_features": 1536,
1195
+ "out_features": 8960,
1196
+ "act_scale": 0.018577444271778498,
1197
+ "smoothquant_alpha": 0.5,
1198
+ "out_dtype": "torch.float16"
1199
+ },
1200
+ "model.language_model.layers.5.mlp.up_proj": {
1201
+ "in_features": 1536,
1202
+ "out_features": 8960,
1203
+ "act_scale": 0.014611972598578986,
1204
+ "smoothquant_alpha": 0.5,
1205
+ "out_dtype": "torch.float16"
1206
+ },
1207
+ "model.language_model.layers.5.mlp.down_proj": {
1208
+ "in_features": 8960,
1209
+ "out_features": 1536,
1210
+ "act_scale": 0.011265405519740788,
1211
+ "smoothquant_alpha": 0.5,
1212
+ "out_dtype": "torch.float16"
1213
+ },
1214
+ "model.language_model.layers.6.self_attn.q_proj": {
1215
+ "in_features": 1536,
1216
+ "out_features": 1536,
1217
+ "act_scale": 0.017126639058270793,
1218
+ "smoothquant_alpha": 0.5,
1219
+ "out_dtype": "torch.float16"
1220
+ },
1221
+ "model.language_model.layers.6.self_attn.k_proj": {
1222
+ "in_features": 1536,
1223
+ "out_features": 256,
1224
+ "act_scale": 0.014979274254145586,
1225
+ "smoothquant_alpha": 0.5,
1226
+ "out_dtype": "torch.float16"
1227
+ },
1228
+ "model.language_model.layers.6.self_attn.v_proj": {
1229
+ "in_features": 1536,
1230
+ "out_features": 256,
1231
+ "act_scale": 0.007247885381142924,
1232
+ "smoothquant_alpha": 0.5,
1233
+ "out_dtype": "torch.float16"
1234
+ },
1235
+ "model.language_model.layers.6.self_attn.o_proj": {
1236
+ "in_features": 1536,
1237
+ "out_features": 1536,
1238
+ "act_scale": 0.0068520542204849365,
1239
+ "smoothquant_alpha": 0.5,
1240
+ "out_dtype": "torch.float16"
1241
+ },
1242
+ "model.language_model.layers.6.mlp.gate_proj": {
1243
+ "in_features": 1536,
1244
+ "out_features": 8960,
1245
+ "act_scale": 0.0188901424407959,
1246
+ "smoothquant_alpha": 0.5,
1247
+ "out_dtype": "torch.float16"
1248
+ },
1249
+ "model.language_model.layers.6.mlp.up_proj": {
1250
+ "in_features": 1536,
1251
+ "out_features": 8960,
1252
+ "act_scale": 0.0178241279181533,
1253
+ "smoothquant_alpha": 0.5,
1254
+ "out_dtype": "torch.float16"
1255
+ },
1256
+ "model.language_model.layers.6.mlp.down_proj": {
1257
+ "in_features": 8960,
1258
+ "out_features": 1536,
1259
+ "act_scale": 0.01482172275152732,
1260
+ "smoothquant_alpha": 0.5,
1261
+ "out_dtype": "torch.float16"
1262
+ },
1263
+ "model.language_model.layers.7.self_attn.q_proj": {
1264
+ "in_features": 1536,
1265
+ "out_features": 1536,
1266
+ "act_scale": 0.01707872255580632,
1267
+ "smoothquant_alpha": 0.5,
1268
+ "out_dtype": "torch.float16"
1269
+ },
1270
+ "model.language_model.layers.7.self_attn.k_proj": {
1271
+ "in_features": 1536,
1272
+ "out_features": 256,
1273
+ "act_scale": 0.012455570416187677,
1274
+ "smoothquant_alpha": 0.5,
1275
+ "out_dtype": "torch.float16"
1276
+ },
1277
+ "model.language_model.layers.7.self_attn.v_proj": {
1278
+ "in_features": 1536,
1279
+ "out_features": 256,
1280
+ "act_scale": 0.011186249612823246,
1281
+ "smoothquant_alpha": 0.5,
1282
+ "out_dtype": "torch.float16"
1283
+ },
1284
+ "model.language_model.layers.7.self_attn.o_proj": {
1285
+ "in_features": 1536,
1286
+ "out_features": 1536,
1287
+ "act_scale": 0.007465832346067654,
1288
+ "smoothquant_alpha": 0.5,
1289
+ "out_dtype": "torch.float16"
1290
+ },
1291
+ "model.language_model.layers.7.mlp.gate_proj": {
1292
+ "in_features": 1536,
1293
+ "out_features": 8960,
1294
+ "act_scale": 0.017271993667121947,
1295
+ "smoothquant_alpha": 0.5,
1296
+ "out_dtype": "torch.float16"
1297
+ },
1298
+ "model.language_model.layers.7.mlp.up_proj": {
1299
+ "in_features": 1536,
1300
+ "out_features": 8960,
1301
+ "act_scale": 0.012164750437098226,
1302
+ "smoothquant_alpha": 0.5,
1303
+ "out_dtype": "torch.float16"
1304
+ },
1305
+ "model.language_model.layers.7.mlp.down_proj": {
1306
+ "in_features": 8960,
1307
+ "out_features": 1536,
1308
+ "act_scale": 0.02268905527009739,
1309
+ "smoothquant_alpha": 0.5,
1310
+ "out_dtype": "torch.float16"
1311
+ },
1312
+ "model.language_model.layers.8.self_attn.q_proj": {
1313
+ "in_features": 1536,
1314
+ "out_features": 1536,
1315
+ "act_scale": 0.01781220698919822,
1316
+ "smoothquant_alpha": 0.5,
1317
+ "out_dtype": "torch.float16"
1318
+ },
1319
+ "model.language_model.layers.8.self_attn.k_proj": {
1320
+ "in_features": 1536,
1321
+ "out_features": 256,
1322
+ "act_scale": 0.011503820344219057,
1323
+ "smoothquant_alpha": 0.5,
1324
+ "out_dtype": "torch.float16"
1325
+ },
1326
+ "model.language_model.layers.8.self_attn.v_proj": {
1327
+ "in_features": 1536,
1328
+ "out_features": 256,
1329
+ "act_scale": 0.01285932852527288,
1330
+ "smoothquant_alpha": 0.5,
1331
+ "out_dtype": "torch.float16"
1332
+ },
1333
+ "model.language_model.layers.8.self_attn.o_proj": {
1334
+ "in_features": 1536,
1335
+ "out_features": 1536,
1336
+ "act_scale": 0.007614903093322994,
1337
+ "smoothquant_alpha": 0.5,
1338
+ "out_dtype": "torch.float16"
1339
+ },
1340
+ "model.language_model.layers.8.mlp.gate_proj": {
1341
+ "in_features": 1536,
1342
+ "out_features": 8960,
1343
+ "act_scale": 0.017289646028533695,
1344
+ "smoothquant_alpha": 0.5,
1345
+ "out_dtype": "torch.float16"
1346
+ },
1347
+ "model.language_model.layers.8.mlp.up_proj": {
1348
+ "in_features": 1536,
1349
+ "out_features": 8960,
1350
+ "act_scale": 0.013838967000405619,
1351
+ "smoothquant_alpha": 0.5,
1352
+ "out_dtype": "torch.float16"
1353
+ },
1354
+ "model.language_model.layers.8.mlp.down_proj": {
1355
+ "in_features": 8960,
1356
+ "out_features": 1536,
1357
+ "act_scale": 0.03071435417715959,
1358
+ "smoothquant_alpha": 0.5,
1359
+ "out_dtype": "torch.float16"
1360
+ },
1361
+ "model.language_model.layers.9.self_attn.q_proj": {
1362
+ "in_features": 1536,
1363
+ "out_features": 1536,
1364
+ "act_scale": 0.0197468404694805,
1365
+ "smoothquant_alpha": 0.5,
1366
+ "out_dtype": "torch.float16"
1367
+ },
1368
+ "model.language_model.layers.9.self_attn.k_proj": {
1369
+ "in_features": 1536,
1370
+ "out_features": 256,
1371
+ "act_scale": 0.016635203924704724,
1372
+ "smoothquant_alpha": 0.5,
1373
+ "out_dtype": "torch.float16"
1374
+ },
1375
+ "model.language_model.layers.9.self_attn.v_proj": {
1376
+ "in_features": 1536,
1377
+ "out_features": 256,
1378
+ "act_scale": 0.009908878897118755,
1379
+ "smoothquant_alpha": 0.5,
1380
+ "out_dtype": "torch.float16"
1381
+ },
1382
+ "model.language_model.layers.9.self_attn.o_proj": {
1383
+ "in_features": 1536,
1384
+ "out_features": 1536,
1385
+ "act_scale": 0.009123546870674674,
1386
+ "smoothquant_alpha": 0.5,
1387
+ "out_dtype": "torch.float16"
1388
+ },
1389
+ "model.language_model.layers.9.mlp.gate_proj": {
1390
+ "in_features": 1536,
1391
+ "out_features": 8960,
1392
+ "act_scale": 0.017440991138848734,
1393
+ "smoothquant_alpha": 0.5,
1394
+ "out_dtype": "torch.float16"
1395
+ },
1396
+ "model.language_model.layers.9.mlp.up_proj": {
1397
+ "in_features": 1536,
1398
+ "out_features": 8960,
1399
+ "act_scale": 0.015315322425421767,
1400
+ "smoothquant_alpha": 0.5,
1401
+ "out_dtype": "torch.float16"
1402
+ },
1403
+ "model.language_model.layers.9.mlp.down_proj": {
1404
+ "in_features": 8960,
1405
+ "out_features": 1536,
1406
+ "act_scale": 0.01320920966741607,
1407
+ "smoothquant_alpha": 0.5,
1408
+ "out_dtype": "torch.float16"
1409
+ },
1410
+ "model.language_model.layers.10.self_attn.q_proj": {
1411
+ "in_features": 1536,
1412
+ "out_features": 1536,
1413
+ "act_scale": 0.019672572143434538,
1414
+ "smoothquant_alpha": 0.5,
1415
+ "out_dtype": "torch.float16"
1416
+ },
1417
+ "model.language_model.layers.10.self_attn.k_proj": {
1418
+ "in_features": 1536,
1419
+ "out_features": 256,
1420
+ "act_scale": 0.015871153103085015,
1421
+ "smoothquant_alpha": 0.5,
1422
+ "out_dtype": "torch.float16"
1423
+ },
1424
+ "model.language_model.layers.10.self_attn.v_proj": {
1425
+ "in_features": 1536,
1426
+ "out_features": 256,
1427
+ "act_scale": 0.00930871738223579,
1428
+ "smoothquant_alpha": 0.5,
1429
+ "out_dtype": "torch.float16"
1430
+ },
1431
+ "model.language_model.layers.10.self_attn.o_proj": {
1432
+ "in_features": 1536,
1433
+ "out_features": 1536,
1434
+ "act_scale": 0.008404247404083492,
1435
+ "smoothquant_alpha": 0.5,
1436
+ "out_dtype": "torch.float16"
1437
+ },
1438
+ "model.language_model.layers.10.mlp.gate_proj": {
1439
+ "in_features": 1536,
1440
+ "out_features": 8960,
1441
+ "act_scale": 0.014881523575369768,
1442
+ "smoothquant_alpha": 0.5,
1443
+ "out_dtype": "torch.float16"
1444
+ },
1445
+ "model.language_model.layers.10.mlp.up_proj": {
1446
+ "in_features": 1536,
1447
+ "out_features": 8960,
1448
+ "act_scale": 0.014982253547728532,
1449
+ "smoothquant_alpha": 0.5,
1450
+ "out_dtype": "torch.float16"
1451
+ },
1452
+ "model.language_model.layers.10.mlp.down_proj": {
1453
+ "in_features": 8960,
1454
+ "out_features": 1536,
1455
+ "act_scale": 0.0114670991897583,
1456
+ "smoothquant_alpha": 0.5,
1457
+ "out_dtype": "torch.float16"
1458
+ },
1459
+ "model.language_model.layers.11.self_attn.q_proj": {
1460
+ "in_features": 1536,
1461
+ "out_features": 1536,
1462
+ "act_scale": 0.01762019555399737,
1463
+ "smoothquant_alpha": 0.5,
1464
+ "out_dtype": "torch.float16"
1465
+ },
1466
+ "model.language_model.layers.11.self_attn.k_proj": {
1467
+ "in_features": 1536,
1468
+ "out_features": 256,
1469
+ "act_scale": 0.016357397469948597,
1470
+ "smoothquant_alpha": 0.5,
1471
+ "out_dtype": "torch.float16"
1472
+ },
1473
+ "model.language_model.layers.11.self_attn.v_proj": {
1474
+ "in_features": 1536,
1475
+ "out_features": 256,
1476
+ "act_scale": 0.010812059162169929,
1477
+ "smoothquant_alpha": 0.5,
1478
+ "out_dtype": "torch.float16"
1479
+ },
1480
+ "model.language_model.layers.11.self_attn.o_proj": {
1481
+ "in_features": 1536,
1482
+ "out_features": 1536,
1483
+ "act_scale": 0.008168115390567329,
1484
+ "smoothquant_alpha": 0.5,
1485
+ "out_dtype": "torch.float16"
1486
+ },
1487
+ "model.language_model.layers.11.mlp.gate_proj": {
1488
+ "in_features": 1536,
1489
+ "out_features": 8960,
1490
+ "act_scale": 0.014244061755383109,
1491
+ "smoothquant_alpha": 0.5,
1492
+ "out_dtype": "torch.float16"
1493
+ },
1494
+ "model.language_model.layers.11.mlp.up_proj": {
1495
+ "in_features": 1536,
1496
+ "out_features": 8960,
1497
+ "act_scale": 0.010984372904920202,
1498
+ "smoothquant_alpha": 0.5,
1499
+ "out_dtype": "torch.float16"
1500
+ },
1501
+ "model.language_model.layers.11.mlp.down_proj": {
1502
+ "in_features": 8960,
1503
+ "out_features": 1536,
1504
+ "act_scale": 0.012976615447697676,
1505
+ "smoothquant_alpha": 0.5,
1506
+ "out_dtype": "torch.float16"
1507
+ },
1508
+ "model.language_model.layers.12.self_attn.q_proj": {
1509
+ "in_features": 1536,
1510
+ "out_features": 1536,
1511
+ "act_scale": 0.015788528862900622,
1512
+ "smoothquant_alpha": 0.5,
1513
+ "out_dtype": "torch.float16"
1514
+ },
1515
+ "model.language_model.layers.12.self_attn.k_proj": {
1516
+ "in_features": 1536,
1517
+ "out_features": 256,
1518
+ "act_scale": 0.016916130471417285,
1519
+ "smoothquant_alpha": 0.5,
1520
+ "out_dtype": "torch.float16"
1521
+ },
1522
+ "model.language_model.layers.12.self_attn.v_proj": {
1523
+ "in_features": 1536,
1524
+ "out_features": 256,
1525
+ "act_scale": 0.009433514489902286,
1526
+ "smoothquant_alpha": 0.5,
1527
+ "out_dtype": "torch.float16"
1528
+ },
1529
+ "model.language_model.layers.12.self_attn.o_proj": {
1530
+ "in_features": 1536,
1531
+ "out_features": 1536,
1532
+ "act_scale": 0.009469956863583543,
1533
+ "smoothquant_alpha": 0.5,
1534
+ "out_dtype": "torch.float16"
1535
+ },
1536
+ "model.language_model.layers.12.mlp.gate_proj": {
1537
+ "in_features": 1536,
1538
+ "out_features": 8960,
1539
+ "act_scale": 0.014436105104881948,
1540
+ "smoothquant_alpha": 0.5,
1541
+ "out_dtype": "torch.float16"
1542
+ },
1543
+ "model.language_model.layers.12.mlp.up_proj": {
1544
+ "in_features": 1536,
1545
+ "out_features": 8960,
1546
+ "act_scale": 0.014943877543051412,
1547
+ "smoothquant_alpha": 0.5,
1548
+ "out_dtype": "torch.float16"
1549
+ },
1550
+ "model.language_model.layers.12.mlp.down_proj": {
1551
+ "in_features": 8960,
1552
+ "out_features": 1536,
1553
+ "act_scale": 0.010102545182535968,
1554
+ "smoothquant_alpha": 0.5,
1555
+ "out_dtype": "torch.float16"
1556
+ },
1557
+ "model.language_model.layers.13.self_attn.q_proj": {
1558
+ "in_features": 1536,
1559
+ "out_features": 1536,
1560
+ "act_scale": 0.017111718185304655,
1561
+ "smoothquant_alpha": 0.5,
1562
+ "out_dtype": "torch.float16"
1563
+ },
1564
+ "model.language_model.layers.13.self_attn.k_proj": {
1565
+ "in_features": 1536,
1566
+ "out_features": 256,
1567
+ "act_scale": 0.016115950787161278,
1568
+ "smoothquant_alpha": 0.5,
1569
+ "out_dtype": "torch.float16"
1570
+ },
1571
+ "model.language_model.layers.13.self_attn.v_proj": {
1572
+ "in_features": 1536,
1573
+ "out_features": 256,
1574
+ "act_scale": 0.008848630537198284,
1575
+ "smoothquant_alpha": 0.5,
1576
+ "out_dtype": "torch.float16"
1577
+ },
1578
+ "model.language_model.layers.13.self_attn.o_proj": {
1579
+ "in_features": 1536,
1580
+ "out_features": 1536,
1581
+ "act_scale": 0.008366376396239274,
1582
+ "smoothquant_alpha": 0.5,
1583
+ "out_dtype": "torch.float16"
1584
+ },
1585
+ "model.language_model.layers.13.mlp.gate_proj": {
1586
+ "in_features": 1536,
1587
+ "out_features": 8960,
1588
+ "act_scale": 0.013314998994662068,
1589
+ "smoothquant_alpha": 0.5,
1590
+ "out_dtype": "torch.float16"
1591
+ },
1592
+ "model.language_model.layers.13.mlp.up_proj": {
1593
+ "in_features": 1536,
1594
+ "out_features": 8960,
1595
+ "act_scale": 0.013658712229390782,
1596
+ "smoothquant_alpha": 0.5,
1597
+ "out_dtype": "torch.float16"
1598
+ },
1599
+ "model.language_model.layers.13.mlp.down_proj": {
1600
+ "in_features": 8960,
1601
+ "out_features": 1536,
1602
+ "act_scale": 0.012686569859662393,
1603
+ "smoothquant_alpha": 0.5,
1604
+ "out_dtype": "torch.float16"
1605
+ },
1606
+ "model.language_model.layers.14.self_attn.q_proj": {
1607
+ "in_features": 1536,
1608
+ "out_features": 1536,
1609
+ "act_scale": 0.015640403342059277,
1610
+ "smoothquant_alpha": 0.5,
1611
+ "out_dtype": "torch.float16"
1612
+ },
1613
+ "model.language_model.layers.14.self_attn.k_proj": {
1614
+ "in_features": 1536,
1615
+ "out_features": 256,
1616
+ "act_scale": 0.01659770274725486,
1617
+ "smoothquant_alpha": 0.5,
1618
+ "out_dtype": "torch.float16"
1619
+ },
1620
+ "model.language_model.layers.14.self_attn.v_proj": {
1621
+ "in_features": 1536,
1622
+ "out_features": 256,
1623
+ "act_scale": 0.01202462789580578,
1624
+ "smoothquant_alpha": 0.5,
1625
+ "out_dtype": "torch.float16"
1626
+ },
1627
+ "model.language_model.layers.14.self_attn.o_proj": {
1628
+ "in_features": 1536,
1629
+ "out_features": 1536,
1630
+ "act_scale": 0.006960612582409475,
1631
+ "smoothquant_alpha": 0.5,
1632
+ "out_dtype": "torch.float16"
1633
+ },
1634
+ "model.language_model.layers.14.mlp.gate_proj": {
1635
+ "in_features": 1536,
1636
+ "out_features": 8960,
1637
+ "act_scale": 0.014151355413001353,
1638
+ "smoothquant_alpha": 0.5,
1639
+ "out_dtype": "torch.float16"
1640
+ },
1641
+ "model.language_model.layers.14.mlp.up_proj": {
1642
+ "in_features": 1536,
1643
+ "out_features": 8960,
1644
+ "act_scale": 0.01234702707275631,
1645
+ "smoothquant_alpha": 0.5,
1646
+ "out_dtype": "torch.float16"
1647
+ },
1648
+ "model.language_model.layers.14.mlp.down_proj": {
1649
+ "in_features": 8960,
1650
+ "out_features": 1536,
1651
+ "act_scale": 0.015763763367660402,
1652
+ "smoothquant_alpha": 0.5,
1653
+ "out_dtype": "torch.float16"
1654
+ },
1655
+ "model.language_model.layers.15.self_attn.q_proj": {
1656
+ "in_features": 1536,
1657
+ "out_features": 1536,
1658
+ "act_scale": 0.017084838837150514,
1659
+ "smoothquant_alpha": 0.5,
1660
+ "out_dtype": "torch.float16"
1661
+ },
1662
+ "model.language_model.layers.15.self_attn.k_proj": {
1663
+ "in_features": 1536,
1664
+ "out_features": 256,
1665
+ "act_scale": 0.016725386221577803,
1666
+ "smoothquant_alpha": 0.5,
1667
+ "out_dtype": "torch.float16"
1668
+ },
1669
+ "model.language_model.layers.15.self_attn.v_proj": {
1670
+ "in_features": 1536,
1671
+ "out_features": 256,
1672
+ "act_scale": 0.008186301847142497,
1673
+ "smoothquant_alpha": 0.5,
1674
+ "out_dtype": "torch.float16"
1675
+ },
1676
+ "model.language_model.layers.15.self_attn.o_proj": {
1677
+ "in_features": 1536,
1678
+ "out_features": 1536,
1679
+ "act_scale": 0.007004190617658961,
1680
+ "smoothquant_alpha": 0.5,
1681
+ "out_dtype": "torch.float16"
1682
+ },
1683
+ "model.language_model.layers.15.mlp.gate_proj": {
1684
+ "in_features": 1536,
1685
+ "out_features": 8960,
1686
+ "act_scale": 0.020452302271925557,
1687
+ "smoothquant_alpha": 0.5,
1688
+ "out_dtype": "torch.float16"
1689
+ },
1690
+ "model.language_model.layers.15.mlp.up_proj": {
1691
+ "in_features": 1536,
1692
+ "out_features": 8960,
1693
+ "act_scale": 0.01974195945919968,
1694
+ "smoothquant_alpha": 0.5,
1695
+ "out_dtype": "torch.float16"
1696
+ },
1697
+ "model.language_model.layers.15.mlp.down_proj": {
1698
+ "in_features": 8960,
1699
+ "out_features": 1536,
1700
+ "act_scale": 0.01462966156756784,
1701
+ "smoothquant_alpha": 0.5,
1702
+ "out_dtype": "torch.float16"
1703
+ },
1704
+ "model.language_model.layers.16.self_attn.q_proj": {
1705
+ "in_features": 1536,
1706
+ "out_features": 1536,
1707
+ "act_scale": 0.02654338633920264,
1708
+ "smoothquant_alpha": 0.5,
1709
+ "out_dtype": "torch.float16"
1710
+ },
1711
+ "model.language_model.layers.16.self_attn.k_proj": {
1712
+ "in_features": 1536,
1713
+ "out_features": 256,
1714
+ "act_scale": 0.017971128929318406,
1715
+ "smoothquant_alpha": 0.5,
1716
+ "out_dtype": "torch.float16"
1717
+ },
1718
+ "model.language_model.layers.16.self_attn.v_proj": {
1719
+ "in_features": 1536,
1720
+ "out_features": 256,
1721
+ "act_scale": 0.007890122143302376,
1722
+ "smoothquant_alpha": 0.5,
1723
+ "out_dtype": "torch.float16"
1724
+ },
1725
+ "model.language_model.layers.16.self_attn.o_proj": {
1726
+ "in_features": 1536,
1727
+ "out_features": 1536,
1728
+ "act_scale": 0.009483855540358176,
1729
+ "smoothquant_alpha": 0.5,
1730
+ "out_dtype": "torch.float16"
1731
+ },
1732
+ "model.language_model.layers.16.mlp.gate_proj": {
1733
+ "in_features": 1536,
1734
+ "out_features": 8960,
1735
+ "act_scale": 0.015570687496756006,
1736
+ "smoothquant_alpha": 0.5,
1737
+ "out_dtype": "torch.float16"
1738
+ },
1739
+ "model.language_model.layers.16.mlp.up_proj": {
1740
+ "in_features": 1536,
1741
+ "out_features": 8960,
1742
+ "act_scale": 0.014243651562788355,
1743
+ "smoothquant_alpha": 0.5,
1744
+ "out_dtype": "torch.float16"
1745
+ },
1746
+ "model.language_model.layers.16.mlp.down_proj": {
1747
+ "in_features": 8960,
1748
+ "out_features": 1536,
1749
+ "act_scale": 0.014674472057913232,
1750
+ "smoothquant_alpha": 0.5,
1751
+ "out_dtype": "torch.float16"
1752
+ },
1753
+ "model.language_model.layers.17.self_attn.q_proj": {
1754
+ "in_features": 1536,
1755
+ "out_features": 1536,
1756
+ "act_scale": 0.017570152057437446,
1757
+ "smoothquant_alpha": 0.5,
1758
+ "out_dtype": "torch.float16"
1759
+ },
1760
+ "model.language_model.layers.17.self_attn.k_proj": {
1761
+ "in_features": 1536,
1762
+ "out_features": 256,
1763
+ "act_scale": 0.02153455366299847,
1764
+ "smoothquant_alpha": 0.5,
1765
+ "out_dtype": "torch.float16"
1766
+ },
1767
+ "model.language_model.layers.17.self_attn.v_proj": {
1768
+ "in_features": 1536,
1769
+ "out_features": 256,
1770
+ "act_scale": 0.012271659580741341,
1771
+ "smoothquant_alpha": 0.5,
1772
+ "out_dtype": "torch.float16"
1773
+ },
1774
+ "model.language_model.layers.17.self_attn.o_proj": {
1775
+ "in_features": 1536,
1776
+ "out_features": 1536,
1777
+ "act_scale": 0.007414881169326662,
1778
+ "smoothquant_alpha": 0.5,
1779
+ "out_dtype": "torch.float16"
1780
+ },
1781
+ "model.language_model.layers.17.mlp.gate_proj": {
1782
+ "in_features": 1536,
1783
+ "out_features": 8960,
1784
+ "act_scale": 0.019823444171214667,
1785
+ "smoothquant_alpha": 0.5,
1786
+ "out_dtype": "torch.float16"
1787
+ },
1788
+ "model.language_model.layers.17.mlp.up_proj": {
1789
+ "in_features": 1536,
1790
+ "out_features": 8960,
1791
+ "act_scale": 0.017401490624495378,
1792
+ "smoothquant_alpha": 0.5,
1793
+ "out_dtype": "torch.float16"
1794
+ },
1795
+ "model.language_model.layers.17.mlp.down_proj": {
1796
+ "in_features": 8960,
1797
+ "out_features": 1536,
1798
+ "act_scale": 0.013188638086394063,
1799
+ "smoothquant_alpha": 0.5,
1800
+ "out_dtype": "torch.float16"
1801
+ },
1802
+ "model.language_model.layers.18.self_attn.q_proj": {
1803
+ "in_features": 1536,
1804
+ "out_features": 1536,
1805
+ "act_scale": 0.018093604741134042,
1806
+ "smoothquant_alpha": 0.5,
1807
+ "out_dtype": "torch.float16"
1808
+ },
1809
+ "model.language_model.layers.18.self_attn.k_proj": {
1810
+ "in_features": 1536,
1811
+ "out_features": 256,
1812
+ "act_scale": 0.01484214977955255,
1813
+ "smoothquant_alpha": 0.5,
1814
+ "out_dtype": "torch.float16"
1815
+ },
1816
+ "model.language_model.layers.18.self_attn.v_proj": {
1817
+ "in_features": 1536,
1818
+ "out_features": 256,
1819
+ "act_scale": 0.012393878200861413,
1820
+ "smoothquant_alpha": 0.5,
1821
+ "out_dtype": "torch.float16"
1822
+ },
1823
+ "model.language_model.layers.18.self_attn.o_proj": {
1824
+ "in_features": 1536,
1825
+ "out_features": 1536,
1826
+ "act_scale": 0.007763971963266688,
1827
+ "smoothquant_alpha": 0.5,
1828
+ "out_dtype": "torch.float16"
1829
+ },
1830
+ "model.language_model.layers.18.mlp.gate_proj": {
1831
+ "in_features": 1536,
1832
+ "out_features": 8960,
1833
+ "act_scale": 0.018470108978391634,
1834
+ "smoothquant_alpha": 0.5,
1835
+ "out_dtype": "torch.float16"
1836
+ },
1837
+ "model.language_model.layers.18.mlp.up_proj": {
1838
+ "in_features": 1536,
1839
+ "out_features": 8960,
1840
+ "act_scale": 0.012738921510891652,
1841
+ "smoothquant_alpha": 0.5,
1842
+ "out_dtype": "torch.float16"
1843
+ },
1844
+ "model.language_model.layers.18.mlp.down_proj": {
1845
+ "in_features": 8960,
1846
+ "out_features": 1536,
1847
+ "act_scale": 0.016149494591660388,
1848
+ "smoothquant_alpha": 0.5,
1849
+ "out_dtype": "torch.float16"
1850
+ },
1851
+ "model.language_model.layers.19.self_attn.q_proj": {
1852
+ "in_features": 1536,
1853
+ "out_features": 1536,
1854
+ "act_scale": 0.018284270143884375,
1855
+ "smoothquant_alpha": 0.5,
1856
+ "out_dtype": "torch.float16"
1857
+ },
1858
+ "model.language_model.layers.19.self_attn.k_proj": {
1859
+ "in_features": 1536,
1860
+ "out_features": 256,
1861
+ "act_scale": 0.01721131895470807,
1862
+ "smoothquant_alpha": 0.5,
1863
+ "out_dtype": "torch.float16"
1864
+ },
1865
+ "model.language_model.layers.19.self_attn.v_proj": {
1866
+ "in_features": 1536,
1867
+ "out_features": 256,
1868
+ "act_scale": 0.009106295315299447,
1869
+ "smoothquant_alpha": 0.5,
1870
+ "out_dtype": "torch.float16"
1871
+ },
1872
+ "model.language_model.layers.19.self_attn.o_proj": {
1873
+ "in_features": 1536,
1874
+ "out_features": 1536,
1875
+ "act_scale": 0.0067378348252904695,
1876
+ "smoothquant_alpha": 0.5,
1877
+ "out_dtype": "torch.float16"
1878
+ },
1879
+ "model.language_model.layers.19.mlp.gate_proj": {
1880
+ "in_features": 1536,
1881
+ "out_features": 8960,
1882
+ "act_scale": 0.019664180560374823,
1883
+ "smoothquant_alpha": 0.5,
1884
+ "out_dtype": "torch.float16"
1885
+ },
1886
+ "model.language_model.layers.19.mlp.up_proj": {
1887
+ "in_features": 1536,
1888
+ "out_features": 8960,
1889
+ "act_scale": 0.014785750644413505,
1890
+ "smoothquant_alpha": 0.5,
1891
+ "out_dtype": "torch.float16"
1892
+ },
1893
+ "model.language_model.layers.19.mlp.down_proj": {
1894
+ "in_features": 8960,
1895
+ "out_features": 1536,
1896
+ "act_scale": 0.017295666566983923,
1897
+ "smoothquant_alpha": 0.5,
1898
+ "out_dtype": "torch.float16"
1899
+ },
1900
+ "model.language_model.layers.20.self_attn.q_proj": {
1901
+ "in_features": 1536,
1902
+ "out_features": 1536,
1903
+ "act_scale": 0.0167858112515427,
1904
+ "smoothquant_alpha": 0.5,
1905
+ "out_dtype": "torch.float16"
1906
+ },
1907
+ "model.language_model.layers.20.self_attn.k_proj": {
1908
+ "in_features": 1536,
1909
+ "out_features": 256,
1910
+ "act_scale": 0.021686362469290184,
1911
+ "smoothquant_alpha": 0.5,
1912
+ "out_dtype": "torch.float16"
1913
+ },
1914
+ "model.language_model.layers.20.self_attn.v_proj": {
1915
+ "in_features": 1536,
1916
+ "out_features": 256,
1917
+ "act_scale": 0.009510900091937207,
1918
+ "smoothquant_alpha": 0.5,
1919
+ "out_dtype": "torch.float16"
1920
+ },
1921
+ "model.language_model.layers.20.self_attn.o_proj": {
1922
+ "in_features": 1536,
1923
+ "out_features": 1536,
1924
+ "act_scale": 0.006677778687064103,
1925
+ "smoothquant_alpha": 0.5,
1926
+ "out_dtype": "torch.float16"
1927
+ },
1928
+ "model.language_model.layers.20.mlp.gate_proj": {
1929
+ "in_features": 1536,
1930
+ "out_features": 8960,
1931
+ "act_scale": 0.018273759075975792,
1932
+ "smoothquant_alpha": 0.5,
1933
+ "out_dtype": "torch.float16"
1934
+ },
1935
+ "model.language_model.layers.20.mlp.up_proj": {
1936
+ "in_features": 1536,
1937
+ "out_features": 8960,
1938
+ "act_scale": 0.01493667993019885,
1939
+ "smoothquant_alpha": 0.5,
1940
+ "out_dtype": "torch.float16"
1941
+ },
1942
+ "model.language_model.layers.20.mlp.down_proj": {
1943
+ "in_features": 8960,
1944
+ "out_features": 1536,
1945
+ "act_scale": 0.016903460495115267,
1946
+ "smoothquant_alpha": 0.5,
1947
+ "out_dtype": "torch.float16"
1948
+ },
1949
+ "model.language_model.layers.21.self_attn.q_proj": {
1950
+ "in_features": 1536,
1951
+ "out_features": 1536,
1952
+ "act_scale": 0.016047697367630606,
1953
+ "smoothquant_alpha": 0.5,
1954
+ "out_dtype": "torch.float16"
1955
+ },
1956
+ "model.language_model.layers.21.self_attn.k_proj": {
1957
+ "in_features": 1536,
1958
+ "out_features": 256,
1959
+ "act_scale": 0.018048720096978617,
1960
+ "smoothquant_alpha": 0.5,
1961
+ "out_dtype": "torch.float16"
1962
+ },
1963
+ "model.language_model.layers.21.self_attn.v_proj": {
1964
+ "in_features": 1536,
1965
+ "out_features": 256,
1966
+ "act_scale": 0.011401965862184059,
1967
+ "smoothquant_alpha": 0.5,
1968
+ "out_dtype": "torch.float16"
1969
+ },
1970
+ "model.language_model.layers.21.self_attn.o_proj": {
1971
+ "in_features": 1536,
1972
+ "out_features": 1536,
1973
+ "act_scale": 0.007243935986766665,
1974
+ "smoothquant_alpha": 0.5,
1975
+ "out_dtype": "torch.float16"
1976
+ },
1977
+ "model.language_model.layers.21.mlp.gate_proj": {
1978
+ "in_features": 1536,
1979
+ "out_features": 8960,
1980
+ "act_scale": 0.01880555265531765,
1981
+ "smoothquant_alpha": 0.5,
1982
+ "out_dtype": "torch.float16"
1983
+ },
1984
+ "model.language_model.layers.21.mlp.up_proj": {
1985
+ "in_features": 1536,
1986
+ "out_features": 8960,
1987
+ "act_scale": 0.017204089427557516,
1988
+ "smoothquant_alpha": 0.5,
1989
+ "out_dtype": "torch.float16"
1990
+ },
1991
+ "model.language_model.layers.21.mlp.down_proj": {
1992
+ "in_features": 8960,
1993
+ "out_features": 1536,
1994
+ "act_scale": 0.02033349097244383,
1995
+ "smoothquant_alpha": 0.5,
1996
+ "out_dtype": "torch.float16"
1997
+ },
1998
+ "model.language_model.layers.22.self_attn.q_proj": {
1999
+ "in_features": 1536,
2000
+ "out_features": 1536,
2001
+ "act_scale": 0.017547941583348072,
2002
+ "smoothquant_alpha": 0.5,
2003
+ "out_dtype": "torch.float16"
2004
+ },
2005
+ "model.language_model.layers.22.self_attn.k_proj": {
2006
+ "in_features": 1536,
2007
+ "out_features": 256,
2008
+ "act_scale": 0.017770733420304427,
2009
+ "smoothquant_alpha": 0.5,
2010
+ "out_dtype": "torch.float16"
2011
+ },
2012
+ "model.language_model.layers.22.self_attn.v_proj": {
2013
+ "in_features": 1536,
2014
+ "out_features": 256,
2015
+ "act_scale": 0.012843428634283111,
2016
+ "smoothquant_alpha": 0.5,
2017
+ "out_dtype": "torch.float16"
2018
+ },
2019
+ "model.language_model.layers.22.self_attn.o_proj": {
2020
+ "in_features": 1536,
2021
+ "out_features": 1536,
2022
+ "act_scale": 0.009375796543331597,
2023
+ "smoothquant_alpha": 0.5,
2024
+ "out_dtype": "torch.float16"
2025
+ },
2026
+ "model.language_model.layers.22.mlp.gate_proj": {
2027
+ "in_features": 1536,
2028
+ "out_features": 8960,
2029
+ "act_scale": 0.01807523149205005,
2030
+ "smoothquant_alpha": 0.5,
2031
+ "out_dtype": "torch.float16"
2032
+ },
2033
+ "model.language_model.layers.22.mlp.up_proj": {
2034
+ "in_features": 1536,
2035
+ "out_features": 8960,
2036
+ "act_scale": 0.016792299240592896,
2037
+ "smoothquant_alpha": 0.5,
2038
+ "out_dtype": "torch.float16"
2039
+ },
2040
+ "model.language_model.layers.22.mlp.down_proj": {
2041
+ "in_features": 8960,
2042
+ "out_features": 1536,
2043
+ "act_scale": 0.02282524484349048,
2044
+ "smoothquant_alpha": 0.5,
2045
+ "out_dtype": "torch.float16"
2046
+ },
2047
+ "model.language_model.layers.23.self_attn.q_proj": {
2048
+ "in_features": 1536,
2049
+ "out_features": 1536,
2050
+ "act_scale": 0.017924070358276367,
2051
+ "smoothquant_alpha": 0.5,
2052
+ "out_dtype": "torch.float16"
2053
+ },
2054
+ "model.language_model.layers.23.self_attn.k_proj": {
2055
+ "in_features": 1536,
2056
+ "out_features": 256,
2057
+ "act_scale": 0.018123487787922537,
2058
+ "smoothquant_alpha": 0.5,
2059
+ "out_dtype": "torch.float16"
2060
+ },
2061
+ "model.language_model.layers.23.self_attn.v_proj": {
2062
+ "in_features": 1536,
2063
+ "out_features": 256,
2064
+ "act_scale": 0.01064312739635077,
2065
+ "smoothquant_alpha": 0.5,
2066
+ "out_dtype": "torch.float16"
2067
+ },
2068
+ "model.language_model.layers.23.self_attn.o_proj": {
2069
+ "in_features": 1536,
2070
+ "out_features": 1536,
2071
+ "act_scale": 0.008809326201911986,
2072
+ "smoothquant_alpha": 0.5,
2073
+ "out_dtype": "torch.float16"
2074
+ },
2075
+ "model.language_model.layers.23.mlp.gate_proj": {
2076
+ "in_features": 1536,
2077
+ "out_features": 8960,
2078
+ "act_scale": 0.018356037890817238,
2079
+ "smoothquant_alpha": 0.5,
2080
+ "out_dtype": "torch.float16"
2081
+ },
2082
+ "model.language_model.layers.23.mlp.up_proj": {
2083
+ "in_features": 1536,
2084
+ "out_features": 8960,
2085
+ "act_scale": 0.016144844490712084,
2086
+ "smoothquant_alpha": 0.5,
2087
+ "out_dtype": "torch.float16"
2088
+ },
2089
+ "model.language_model.layers.23.mlp.down_proj": {
2090
+ "in_features": 8960,
2091
+ "out_features": 1536,
2092
+ "act_scale": 0.020913634713240495,
2093
+ "smoothquant_alpha": 0.5,
2094
+ "out_dtype": "torch.float16"
2095
+ },
2096
+ "model.language_model.layers.24.self_attn.q_proj": {
2097
+ "in_features": 1536,
2098
+ "out_features": 1536,
2099
+ "act_scale": 0.020576118484256774,
2100
+ "smoothquant_alpha": 0.5,
2101
+ "out_dtype": "torch.float16"
2102
+ },
2103
+ "model.language_model.layers.24.self_attn.k_proj": {
2104
+ "in_features": 1536,
2105
+ "out_features": 256,
2106
+ "act_scale": 0.016015707977174772,
2107
+ "smoothquant_alpha": 0.5,
2108
+ "out_dtype": "torch.float16"
2109
+ },
2110
+ "model.language_model.layers.24.self_attn.v_proj": {
2111
+ "in_features": 1536,
2112
+ "out_features": 256,
2113
+ "act_scale": 0.010398974568825069,
2114
+ "smoothquant_alpha": 0.5,
2115
+ "out_dtype": "torch.float16"
2116
+ },
2117
+ "model.language_model.layers.24.self_attn.o_proj": {
2118
+ "in_features": 1536,
2119
+ "out_features": 1536,
2120
+ "act_scale": 0.006741659378442239,
2121
+ "smoothquant_alpha": 0.5,
2122
+ "out_dtype": "torch.float16"
2123
+ },
2124
+ "model.language_model.layers.24.mlp.gate_proj": {
2125
+ "in_features": 1536,
2126
+ "out_features": 8960,
2127
+ "act_scale": 0.01695194582300862,
2128
+ "smoothquant_alpha": 0.5,
2129
+ "out_dtype": "torch.float16"
2130
+ },
2131
+ "model.language_model.layers.24.mlp.up_proj": {
2132
+ "in_features": 1536,
2133
+ "out_features": 8960,
2134
+ "act_scale": 0.016808357764416793,
2135
+ "smoothquant_alpha": 0.5,
2136
+ "out_dtype": "torch.float16"
2137
+ },
2138
+ "model.language_model.layers.24.mlp.down_proj": {
2139
+ "in_features": 8960,
2140
+ "out_features": 1536,
2141
+ "act_scale": 0.02610544332369106,
2142
+ "smoothquant_alpha": 0.5,
2143
+ "out_dtype": "torch.float16"
2144
+ },
2145
+ "model.language_model.layers.25.self_attn.q_proj": {
2146
+ "in_features": 1536,
2147
+ "out_features": 1536,
2148
+ "act_scale": 0.01670930517001415,
2149
+ "smoothquant_alpha": 0.5,
2150
+ "out_dtype": "torch.float16"
2151
+ },
2152
+ "model.language_model.layers.25.self_attn.k_proj": {
2153
+ "in_features": 1536,
2154
+ "out_features": 256,
2155
+ "act_scale": 0.016933991214421792,
2156
+ "smoothquant_alpha": 0.5,
2157
+ "out_dtype": "torch.float16"
2158
+ },
2159
+ "model.language_model.layers.25.self_attn.v_proj": {
2160
+ "in_features": 1536,
2161
+ "out_features": 256,
2162
+ "act_scale": 0.01482135104382132,
2163
+ "smoothquant_alpha": 0.5,
2164
+ "out_dtype": "torch.float16"
2165
+ },
2166
+ "model.language_model.layers.25.self_attn.o_proj": {
2167
+ "in_features": 1536,
2168
+ "out_features": 1536,
2169
+ "act_scale": 0.010464924526965524,
2170
+ "smoothquant_alpha": 0.5,
2171
+ "out_dtype": "torch.float16"
2172
+ },
2173
+ "model.language_model.layers.25.mlp.gate_proj": {
2174
+ "in_features": 1536,
2175
+ "out_features": 8960,
2176
+ "act_scale": 0.03200663168599287,
2177
+ "smoothquant_alpha": 0.5,
2178
+ "out_dtype": "torch.float16"
2179
+ },
2180
+ "model.language_model.layers.25.mlp.up_proj": {
2181
+ "in_features": 1536,
2182
+ "out_features": 8960,
2183
+ "act_scale": 0.016040488490908163,
2184
+ "smoothquant_alpha": 0.5,
2185
+ "out_dtype": "torch.float16"
2186
+ },
2187
+ "model.language_model.layers.25.mlp.down_proj": {
2188
+ "in_features": 8960,
2189
+ "out_features": 1536,
2190
+ "act_scale": 0.03664320097194882,
2191
+ "smoothquant_alpha": 0.5,
2192
+ "out_dtype": "torch.float16"
2193
+ },
2194
+ "model.language_model.layers.26.self_attn.q_proj": {
2195
+ "in_features": 1536,
2196
+ "out_features": 1536,
2197
+ "act_scale": 0.019067401961078794,
2198
+ "smoothquant_alpha": 0.5,
2199
+ "out_dtype": "torch.float16"
2200
+ },
2201
+ "model.language_model.layers.26.self_attn.k_proj": {
2202
+ "in_features": 1536,
2203
+ "out_features": 256,
2204
+ "act_scale": 0.01596015081630917,
2205
+ "smoothquant_alpha": 0.5,
2206
+ "out_dtype": "torch.float16"
2207
+ },
2208
+ "model.language_model.layers.26.self_attn.v_proj": {
2209
+ "in_features": 1536,
2210
+ "out_features": 256,
2211
+ "act_scale": 0.015354877381812869,
2212
+ "smoothquant_alpha": 0.5,
2213
+ "out_dtype": "torch.float16"
2214
+ },
2215
+ "model.language_model.layers.26.self_attn.o_proj": {
2216
+ "in_features": 1536,
2217
+ "out_features": 1536,
2218
+ "act_scale": 0.010470243889515795,
2219
+ "smoothquant_alpha": 0.5,
2220
+ "out_dtype": "torch.float16"
2221
+ },
2222
+ "model.language_model.layers.26.mlp.gate_proj": {
2223
+ "in_features": 1536,
2224
+ "out_features": 8960,
2225
+ "act_scale": 0.024390772571713907,
2226
+ "smoothquant_alpha": 0.5,
2227
+ "out_dtype": "torch.float16"
2228
+ },
2229
+ "model.language_model.layers.26.mlp.up_proj": {
2230
+ "in_features": 1536,
2231
+ "out_features": 8960,
2232
+ "act_scale": 0.024410642038180132,
2233
+ "smoothquant_alpha": 0.5,
2234
+ "out_dtype": "torch.float16"
2235
+ },
2236
+ "model.language_model.layers.26.mlp.down_proj": {
2237
+ "in_features": 8960,
2238
+ "out_features": 1536,
2239
+ "act_scale": 0.2042921621968427,
2240
+ "smoothquant_alpha": 0.5,
2241
+ "out_dtype": "torch.float16"
2242
+ },
2243
+ "model.language_model.layers.27.self_attn.q_proj": {
2244
+ "in_features": 1536,
2245
+ "out_features": 1536,
2246
+ "act_scale": 0.016146177381981076,
2247
+ "smoothquant_alpha": 0.5,
2248
+ "out_dtype": "torch.float16"
2249
+ },
2250
+ "model.language_model.layers.27.self_attn.k_proj": {
2251
+ "in_features": 1536,
2252
+ "out_features": 256,
2253
+ "act_scale": 0.01712871348764014,
2254
+ "smoothquant_alpha": 0.5,
2255
+ "out_dtype": "torch.float16"
2256
+ },
2257
+ "model.language_model.layers.27.self_attn.v_proj": {
2258
+ "in_features": 1536,
2259
+ "out_features": 256,
2260
+ "act_scale": 0.013760393060098483,
2261
+ "smoothquant_alpha": 0.5,
2262
+ "out_dtype": "torch.float16"
2263
+ },
2264
+ "model.language_model.layers.27.self_attn.o_proj": {
2265
+ "in_features": 1536,
2266
+ "out_features": 1536,
2267
+ "act_scale": 0.014997026113074595,
2268
+ "smoothquant_alpha": 0.5,
2269
+ "out_dtype": "torch.float16"
2270
+ },
2271
+ "model.language_model.layers.27.mlp.gate_proj": {
2272
+ "in_features": 1536,
2273
+ "out_features": 8960,
2274
+ "act_scale": 0.026463501096710446,
2275
+ "smoothquant_alpha": 0.5,
2276
+ "out_dtype": "torch.float16"
2277
+ },
2278
+ "model.language_model.layers.27.mlp.up_proj": {
2279
+ "in_features": 1536,
2280
+ "out_features": 8960,
2281
+ "act_scale": 0.03419333931029312,
2282
+ "smoothquant_alpha": 0.5,
2283
+ "out_dtype": "torch.float16"
2284
+ },
2285
+ "model.language_model.layers.27.mlp.down_proj": {
2286
+ "in_features": 8960,
2287
+ "out_features": 1536,
2288
+ "act_scale": 0.07352166663943313,
2289
+ "smoothquant_alpha": 0.5,
2290
+ "out_dtype": "torch.float16"
2291
+ },
2292
+ "lm_head": {
2293
+ "in_features": 1536,
2294
+ "out_features": 151936,
2295
+ "act_scale": 0.019375630251065954,
2296
+ "smoothquant_alpha": 0.5,
2297
+ "out_dtype": "torch.float16"
2298
+ }
2299
+ },
2300
+ "self_contained_bundle": true,
2301
+ "requires_base_model_download": false
2302
+ }
w8a8_smoothquant_state.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41e1d7461c85307d17ac88ca6bd282a39ef0e2934922fd1d4721c04d5f0e35e8
3
+ size 2687371459