Hugging Face's logo

kkeezz
/
cap-iaa-lora-attn

PEFT
mplug_owl2
Model card Files Community
kkeezz commited on
Commit
0b30d8f
·
verified ·
1 Parent(s): e6f535a

Upload 5 files

Browse files
Files changed (5) hide show
  1. README.md +9 -0
  2. adapter_config.json +83 -0
  3. config.json +184 -0
  4. preprocessor_config.json +20 -0
  5. trainer_state.json +0 -0
README.md ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+ ### Framework versions
7
+
8
+
9
+ - PEFT 0.5.0
adapter_config.json ADDED
@@ -0,0 +1,83 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "MAGAer13/mplug-owl2-llama2-7b",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 256,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 128,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "model.layers.18.self_attn.v_proj.multiway.1",
18
+ "model.layers.14.self_attn.v_proj.multiway.1",
19
+ "model.layers.27.self_attn.q_proj",
20
+ "model.layers.2.self_attn.q_proj",
21
+ "model.layers.4.self_attn.q_proj",
22
+ "model.layers.18.self_attn.q_proj",
23
+ "model.layers.7.self_attn.q_proj",
24
+ "model.layers.5.self_attn.q_proj",
25
+ "model.layers.15.self_attn.q_proj",
26
+ "model.layers.7.self_attn.v_proj.multiway.1",
27
+ "model.layers.6.self_attn.v_proj.multiway.1",
28
+ "model.layers.12.self_attn.v_proj.multiway.1",
29
+ "model.layers.16.self_attn.q_proj",
30
+ "model.layers.31.self_attn.q_proj",
31
+ "model.layers.17.self_attn.q_proj",
32
+ "model.layers.10.self_attn.q_proj",
33
+ "model.layers.22.self_attn.q_proj",
34
+ "model.layers.3.self_attn.v_proj.multiway.1",
35
+ "model.layers.23.self_attn.q_proj",
36
+ "model.layers.9.self_attn.v_proj.multiway.1",
37
+ "model.layers.13.self_attn.q_proj",
38
+ "model.layers.4.self_attn.v_proj.multiway.1",
39
+ "model.layers.11.self_attn.v_proj.multiway.1",
40
+ "model.layers.19.self_attn.v_proj.multiway.1",
41
+ "model.layers.21.self_attn.q_proj",
42
+ "model.layers.26.self_attn.q_proj",
43
+ "model.layers.28.self_attn.v_proj.multiway.1",
44
+ "model.layers.20.self_attn.q_proj",
45
+ "model.layers.31.self_attn.v_proj.multiway.1",
46
+ "model.layers.0.self_attn.q_proj",
47
+ "model.layers.24.self_attn.v_proj.multiway.1",
48
+ "model.layers.30.self_attn.q_proj",
49
+ "model.layers.5.self_attn.v_proj.multiway.1",
50
+ "model.layers.17.self_attn.v_proj.multiway.1",
51
+ "model.layers.27.self_attn.v_proj.multiway.1",
52
+ "model.layers.13.self_attn.v_proj.multiway.1",
53
+ "model.layers.6.self_attn.q_proj",
54
+ "model.layers.12.self_attn.q_proj",
55
+ "model.layers.25.self_attn.q_proj",
56
+ "model.layers.2.self_attn.v_proj.multiway.1",
57
+ "model.layers.1.self_attn.q_proj",
58
+ "model.layers.10.self_attn.v_proj.multiway.1",
59
+ "model.layers.29.self_attn.v_proj.multiway.1",
60
+ "model.layers.30.self_attn.v_proj.multiway.1",
61
+ "model.layers.3.self_attn.q_proj",
62
+ "model.layers.26.self_attn.v_proj.multiway.1",
63
+ "model.layers.8.self_attn.v_proj.multiway.1",
64
+ "model.layers.8.self_attn.q_proj",
65
+ "model.layers.11.self_attn.q_proj",
66
+ "model.layers.20.self_attn.v_proj.multiway.1",
67
+ "model.layers.24.self_attn.q_proj",
68
+ "model.layers.23.self_attn.v_proj.multiway.1",
69
+ "model.layers.28.self_attn.q_proj",
70
+ "model.layers.9.self_attn.q_proj",
71
+ "model.layers.14.self_attn.q_proj",
72
+ "model.layers.22.self_attn.v_proj.multiway.1",
73
+ "model.layers.19.self_attn.q_proj",
74
+ "model.layers.21.self_attn.v_proj.multiway.1",
75
+ "model.layers.15.self_attn.v_proj.multiway.1",
76
+ "model.layers.29.self_attn.q_proj",
77
+ "model.layers.16.self_attn.v_proj.multiway.1",
78
+ "model.layers.0.self_attn.v_proj.multiway.1",
79
+ "model.layers.25.self_attn.v_proj.multiway.1",
80
+ "model.layers.1.self_attn.v_proj.multiway.1"
81
+ ],
82
+ "task_type": "CAUSAL_LM"
83
+ }
config.json ADDED
@@ -0,0 +1,184 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "MAGAer13/mplug-owl2-llama2-7b",
3
+ "attention_bias": false,
4
+ "attention_dropout": 0.0,
5
+ "attn_heads": 8,
6
+ "bos_token_id": 1,
7
+ "eos_token_id": 2,
8
+ "freeze_vision_model": false,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 4096,
11
+ "image_aspect_ratio": "pad",
12
+ "image_grid_pinpoints": null,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 11008,
15
+ "max_position_embeddings": 2048,
16
+ "model_type": "mplug_owl2",
17
+ "num_attention_heads": 32,
18
+ "num_hidden_layers": 32,
19
+ "num_key_value_heads": 32,
20
+ "pretraining_tp": 1,
21
+ "rms_norm_eps": 1e-06,
22
+ "rope_scaling": null,
23
+ "rope_theta": 10000.0,
24
+ "tie_word_embeddings": false,
25
+ "transformers_version": "4.36.1",
26
+ "tune_visual_abstractor": true,
27
+ "use_cache": true,
28
+ "visual_abstractor_lr": 2e-05,
29
+ "visual_config": {
30
+ "visual_abstractor": {
31
+ "_name_or_path": "",
32
+ "add_cross_attention": false,
33
+ "architectures": null,
34
+ "attention_probs_dropout_prob": 0.0,
35
+ "bad_words_ids": null,
36
+ "begin_suppress_tokens": null,
37
+ "bos_token_id": null,
38
+ "chunk_size_feed_forward": 0,
39
+ "cross_attention_hidden_size": null,
40
+ "decoder_start_token_id": null,
41
+ "diversity_penalty": 0.0,
42
+ "do_sample": false,
43
+ "early_stopping": false,
44
+ "encoder_hidden_size": 1024,
45
+ "encoder_no_repeat_ngram_size": 0,
46
+ "eos_token_id": null,
47
+ "exponential_decay_length_penalty": null,
48
+ "finetuning_task": null,
49
+ "forced_bos_token_id": null,
50
+ "forced_eos_token_id": null,
51
+ "grid_size": 32,
52
+ "hidden_size": 1024,
53
+ "id2label": {
54
+ "0": "LABEL_0",
55
+ "1": "LABEL_1"
56
+ },
57
+ "initializer_range": 0.02,
58
+ "intermediate_size": 2816,
59
+ "is_decoder": false,
60
+ "is_encoder_decoder": false,
61
+ "label2id": {
62
+ "LABEL_0": 0,
63
+ "LABEL_1": 1
64
+ },
65
+ "layer_norm_eps": 1e-06,
66
+ "length_penalty": 1.0,
67
+ "max_length": 20,
68
+ "min_length": 0,
69
+ "model_type": "mplug_owl_visual_abstract",
70
+ "no_repeat_ngram_size": 0,
71
+ "num_attention_heads": 16,
72
+ "num_beam_groups": 1,
73
+ "num_beams": 1,
74
+ "num_hidden_layers": 6,
75
+ "num_learnable_queries": 64,
76
+ "num_return_sequences": 1,
77
+ "output_attentions": false,
78
+ "output_hidden_states": false,
79
+ "output_scores": false,
80
+ "pad_token_id": null,
81
+ "prefix": null,
82
+ "problem_type": null,
83
+ "pruned_heads": {},
84
+ "remove_invalid_values": false,
85
+ "repetition_penalty": 1.0,
86
+ "return_dict": true,
87
+ "return_dict_in_generate": false,
88
+ "sep_token_id": null,
89
+ "suppress_tokens": null,
90
+ "task_specific_params": null,
91
+ "temperature": 1.0,
92
+ "tf_legacy_loss": false,
93
+ "tie_encoder_decoder": false,
94
+ "tie_word_embeddings": true,
95
+ "tokenizer_class": null,
96
+ "top_k": 50,
97
+ "top_p": 1.0,
98
+ "torch_dtype": null,
99
+ "torchscript": false,
100
+ "transformers_version": "4.28.1",
101
+ "typical_p": 1.0,
102
+ "use_bfloat16": false
103
+ },
104
+ "visual_model": {
105
+ "_name_or_path": "",
106
+ "add_cross_attention": false,
107
+ "architectures": null,
108
+ "attention_dropout": 0.0,
109
+ "bad_words_ids": null,
110
+ "begin_suppress_tokens": null,
111
+ "bos_token_id": null,
112
+ "chunk_size_feed_forward": 0,
113
+ "cross_attention_hidden_size": null,
114
+ "decoder_start_token_id": null,
115
+ "diversity_penalty": 0.0,
116
+ "do_sample": false,
117
+ "early_stopping": false,
118
+ "encoder_no_repeat_ngram_size": 0,
119
+ "eos_token_id": null,
120
+ "exponential_decay_length_penalty": null,
121
+ "finetuning_task": null,
122
+ "forced_bos_token_id": null,
123
+ "forced_eos_token_id": null,
124
+ "hidden_act": "quick_gelu",
125
+ "hidden_size": 1024,
126
+ "id2label": {
127
+ "0": "LABEL_0",
128
+ "1": "LABEL_1"
129
+ },
130
+ "image_size": 448,
131
+ "initializer_factor": 1.0,
132
+ "initializer_range": 0.02,
133
+ "intermediate_size": 4096,
134
+ "is_decoder": false,
135
+ "is_encoder_decoder": false,
136
+ "label2id": {
137
+ "LABEL_0": 0,
138
+ "LABEL_1": 1
139
+ },
140
+ "layer_norm_eps": 1e-06,
141
+ "length_penalty": 1.0,
142
+ "max_length": 20,
143
+ "min_length": 0,
144
+ "model_type": "mplug_owl_vision_model",
145
+ "no_repeat_ngram_size": 0,
146
+ "num_attention_heads": 16,
147
+ "num_beam_groups": 1,
148
+ "num_beams": 1,
149
+ "num_channels": 3,
150
+ "num_hidden_layers": 24,
151
+ "num_return_sequences": 1,
152
+ "output_attentions": false,
153
+ "output_hidden_states": false,
154
+ "output_scores": false,
155
+ "pad_token_id": null,
156
+ "patch_size": 14,
157
+ "prefix": null,
158
+ "problem_type": null,
159
+ "projection_dim": 768,
160
+ "pruned_heads": {},
161
+ "remove_invalid_values": false,
162
+ "repetition_penalty": 1.0,
163
+ "return_dict": true,
164
+ "return_dict_in_generate": false,
165
+ "sep_token_id": null,
166
+ "suppress_tokens": null,
167
+ "task_specific_params": null,
168
+ "temperature": 1.0,
169
+ "tf_legacy_loss": false,
170
+ "tie_encoder_decoder": false,
171
+ "tie_word_embeddings": true,
172
+ "tokenizer_class": null,
173
+ "top_k": 50,
174
+ "top_p": 1.0,
175
+ "torch_dtype": null,
176
+ "torchscript": false,
177
+ "transformers_version": "4.28.1",
178
+ "typical_p": 1.0,
179
+ "use_bfloat16": false,
180
+ "use_flash_attn": false
181
+ }
182
+ },
183
+ "vocab_size": 32000
184
+ }
preprocessor_config.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "crop_size": 448,
3
+ "do_center_crop": true,
4
+ "do_normalize": true,
5
+ "do_resize": true,
6
+ "feature_extractor_type": "CLIPFeatureExtractor",
7
+ "image_mean": [
8
+ 0.48145466,
9
+ 0.4578275,
10
+ 0.40821073
11
+ ],
12
+ "image_std": [
13
+ 0.26862954,
14
+ 0.26130258,
15
+ 0.27577711
16
+ ],
17
+ "resample": 3,
18
+ "size": 448
19
+ }
20
+
trainer_state.json ADDED
The diff for this file is too large to render. See raw diff