Upload folder using huggingface_hub
Browse files- .gitattributes +1 -0
- model_index.json +33 -0
- processor/chat_template.jinja +13 -0
- processor/processor_config.json +31 -0
- processor/tokenizer.json +3 -0
- processor/tokenizer_config.json +44 -0
- scheduler/scheduler_config.json +18 -0
- text_encoder/config.json +32 -0
- text_encoder/model.safetensors +3 -0
- tokenizer/added_tokens.json +127 -0
- tokenizer/tokenizer_config.json +1292 -0
- transformer/config.json +266 -0
- transformer/diffusion_pytorch_model.safetensors +3 -0
- transformer/quantization_config.json +251 -0
- vae/config.json +72 -0
- vae/diffusion_pytorch_model.safetensors +3 -0
- vision_language_encoder/config.json +505 -0
- vision_language_encoder/generation_config.json +10 -0
- vision_language_encoder/model-00001-of-00002.safetensors +3 -0
- vision_language_encoder/model-00002-of-00002.safetensors +3 -0
- vision_language_encoder/model.safetensors.index.json +0 -0
- vision_language_encoder/quantization_config.json +438 -0
.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 |
+
processor/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
model_index.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_class_name": "GlmImagePipeline",
|
| 3 |
+
"_diffusers_version": "0.37.0.dev0",
|
| 4 |
+
"_name_or_path": "zai-org/GLM-Image",
|
| 5 |
+
"processor": [
|
| 6 |
+
"glm_image",
|
| 7 |
+
"GlmImageProcessor"
|
| 8 |
+
],
|
| 9 |
+
"scheduler": [
|
| 10 |
+
"diffusers",
|
| 11 |
+
"FlowMatchEulerDiscreteScheduler"
|
| 12 |
+
],
|
| 13 |
+
"text_encoder": [
|
| 14 |
+
"transformers",
|
| 15 |
+
"T5EncoderModel"
|
| 16 |
+
],
|
| 17 |
+
"tokenizer": [
|
| 18 |
+
"transformers",
|
| 19 |
+
"ByT5Tokenizer"
|
| 20 |
+
],
|
| 21 |
+
"transformer": [
|
| 22 |
+
"diffusers",
|
| 23 |
+
"GlmImageTransformer2DModel"
|
| 24 |
+
],
|
| 25 |
+
"vae": [
|
| 26 |
+
"diffusers",
|
| 27 |
+
"AutoencoderKL"
|
| 28 |
+
],
|
| 29 |
+
"vision_language_encoder": [
|
| 30 |
+
"glm_image",
|
| 31 |
+
"GlmImageForConditionalGeneration"
|
| 32 |
+
]
|
| 33 |
+
}
|
processor/chat_template.jinja
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{%- for m in messages -%}
|
| 2 |
+
{%- if m.content is string -%}
|
| 3 |
+
{{ m.content }}
|
| 4 |
+
{%- else -%}
|
| 5 |
+
{%- for item in m.content -%}
|
| 6 |
+
{%- if item.type == 'image' or item.get('image') is not none -%}
|
| 7 |
+
<|dit_token_16384|><|image|><|dit_token_16385|>
|
| 8 |
+
{%- elif item.type == 'text' -%}
|
| 9 |
+
{{ item.text }}
|
| 10 |
+
{%- endif -%}
|
| 11 |
+
{%- endfor -%}
|
| 12 |
+
{%- endif -%}
|
| 13 |
+
{%- endfor -%}
|
processor/processor_config.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"image_processor": {
|
| 3 |
+
"do_convert_rgb": true,
|
| 4 |
+
"do_normalize": true,
|
| 5 |
+
"do_rescale": true,
|
| 6 |
+
"do_resize": true,
|
| 7 |
+
"image_mean": [
|
| 8 |
+
0.5,
|
| 9 |
+
0.5,
|
| 10 |
+
0.5
|
| 11 |
+
],
|
| 12 |
+
"image_processor_type": "GlmImageImageProcessor",
|
| 13 |
+
"image_std": [
|
| 14 |
+
0.5,
|
| 15 |
+
0.5,
|
| 16 |
+
0.5
|
| 17 |
+
],
|
| 18 |
+
"max_pixels": 4194304,
|
| 19 |
+
"merge_size": 1,
|
| 20 |
+
"min_pixels": 262144,
|
| 21 |
+
"patch_size": 16,
|
| 22 |
+
"resample": 3,
|
| 23 |
+
"rescale_factor": 0.00392156862745098,
|
| 24 |
+
"size": {
|
| 25 |
+
"longest_edge": 4194304,
|
| 26 |
+
"shortest_edge": 262144
|
| 27 |
+
},
|
| 28 |
+
"temporal_patch_size": 1
|
| 29 |
+
},
|
| 30 |
+
"processor_class": "GlmImageProcessor"
|
| 31 |
+
}
|
processor/tokenizer.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d771c9659fd2c65c0ef8a8daf9d9f9f07b53f89ce164cc832a28137db604d01d
|
| 3 |
+
size 23765636
|
processor/tokenizer_config.json
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"backend": "tokenizers",
|
| 3 |
+
"bos_token": "<|dit_token_16384|>",
|
| 4 |
+
"clean_up_tokenization_spaces": false,
|
| 5 |
+
"do_lower_case": false,
|
| 6 |
+
"eos_token": "<|dit_token_16385|>",
|
| 7 |
+
"extra_special_tokens": [
|
| 8 |
+
"<|endoftext|>",
|
| 9 |
+
"[MASK]",
|
| 10 |
+
"[gMASK]",
|
| 11 |
+
"[sMASK]",
|
| 12 |
+
"<sop>",
|
| 13 |
+
"<eop>",
|
| 14 |
+
"<|system|>",
|
| 15 |
+
"<|user|>",
|
| 16 |
+
"<|assistant|>",
|
| 17 |
+
"<|observation|>",
|
| 18 |
+
"<|begin_of_image|>",
|
| 19 |
+
"<|end_of_image|>",
|
| 20 |
+
"<|begin_of_video|>",
|
| 21 |
+
"<|end_of_video|>",
|
| 22 |
+
"<|image|>",
|
| 23 |
+
"<|video|>"
|
| 24 |
+
],
|
| 25 |
+
"grid_bos_token": "<sop>",
|
| 26 |
+
"grid_eos_token": "<eop>",
|
| 27 |
+
"image_token": "<|image|>",
|
| 28 |
+
"is_local": true,
|
| 29 |
+
"model_input_names": [
|
| 30 |
+
"input_ids",
|
| 31 |
+
"attention_mask"
|
| 32 |
+
],
|
| 33 |
+
"model_max_length": 65536,
|
| 34 |
+
"model_specific_special_tokens": {
|
| 35 |
+
"grid_bos_token": "<sop>",
|
| 36 |
+
"grid_eos_token": "<eop>",
|
| 37 |
+
"image_token": "<|image|>"
|
| 38 |
+
},
|
| 39 |
+
"pad_token": "<|dit_token_16385|>",
|
| 40 |
+
"padding_side": "left",
|
| 41 |
+
"processor_class": "GlmImageProcessor",
|
| 42 |
+
"remove_space": false,
|
| 43 |
+
"tokenizer_class": "TokenizersBackend"
|
| 44 |
+
}
|
scheduler/scheduler_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_class_name": "FlowMatchEulerDiscreteScheduler",
|
| 3 |
+
"_diffusers_version": "0.37.0.dev0",
|
| 4 |
+
"base_image_seq_len": 256,
|
| 5 |
+
"base_shift": 0.25,
|
| 6 |
+
"invert_sigmas": false,
|
| 7 |
+
"max_image_seq_len": 4096,
|
| 8 |
+
"max_shift": 0.75,
|
| 9 |
+
"num_train_timesteps": 1000,
|
| 10 |
+
"shift": 1.0,
|
| 11 |
+
"shift_terminal": null,
|
| 12 |
+
"stochastic_sampling": false,
|
| 13 |
+
"time_shift_type": "linear",
|
| 14 |
+
"use_beta_sigmas": false,
|
| 15 |
+
"use_dynamic_shifting": true,
|
| 16 |
+
"use_exponential_sigmas": false,
|
| 17 |
+
"use_karras_sigmas": false
|
| 18 |
+
}
|
text_encoder/config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"T5EncoderModel"
|
| 4 |
+
],
|
| 5 |
+
"classifier_dropout": 0.0,
|
| 6 |
+
"d_ff": 3584,
|
| 7 |
+
"d_kv": 64,
|
| 8 |
+
"d_model": 1472,
|
| 9 |
+
"decoder_start_token_id": 0,
|
| 10 |
+
"dense_act_fn": "gelu_new",
|
| 11 |
+
"dropout_rate": 0.1,
|
| 12 |
+
"dtype": "bfloat16",
|
| 13 |
+
"eos_token_id": 1,
|
| 14 |
+
"feed_forward_proj": "gated-gelu",
|
| 15 |
+
"gradient_checkpointing": false,
|
| 16 |
+
"initializer_factor": 1.0,
|
| 17 |
+
"is_encoder_decoder": false,
|
| 18 |
+
"is_gated_act": true,
|
| 19 |
+
"layer_norm_epsilon": 1e-06,
|
| 20 |
+
"model_type": "t5",
|
| 21 |
+
"num_decoder_layers": 4,
|
| 22 |
+
"num_heads": 6,
|
| 23 |
+
"num_layers": 12,
|
| 24 |
+
"pad_token_id": 0,
|
| 25 |
+
"relative_attention_max_distance": 128,
|
| 26 |
+
"relative_attention_num_buckets": 32,
|
| 27 |
+
"scale_decoder_outputs": false,
|
| 28 |
+
"tokenizer_class": "ByT5Tokenizer",
|
| 29 |
+
"transformers_version": "5.0.0.dev0",
|
| 30 |
+
"use_cache": false,
|
| 31 |
+
"vocab_size": 384
|
| 32 |
+
}
|
text_encoder/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7bc470014c3782064177d29ca80f9b03da30f0e259f6fa06c256bc5ecdf600d6
|
| 3 |
+
size 561943976
|
tokenizer/added_tokens.json
ADDED
|
@@ -0,0 +1,127 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<extra_id_0>": 259,
|
| 3 |
+
"<extra_id_100>": 359,
|
| 4 |
+
"<extra_id_101>": 360,
|
| 5 |
+
"<extra_id_102>": 361,
|
| 6 |
+
"<extra_id_103>": 362,
|
| 7 |
+
"<extra_id_104>": 363,
|
| 8 |
+
"<extra_id_105>": 364,
|
| 9 |
+
"<extra_id_106>": 365,
|
| 10 |
+
"<extra_id_107>": 366,
|
| 11 |
+
"<extra_id_108>": 367,
|
| 12 |
+
"<extra_id_109>": 368,
|
| 13 |
+
"<extra_id_10>": 269,
|
| 14 |
+
"<extra_id_110>": 369,
|
| 15 |
+
"<extra_id_111>": 370,
|
| 16 |
+
"<extra_id_112>": 371,
|
| 17 |
+
"<extra_id_113>": 372,
|
| 18 |
+
"<extra_id_114>": 373,
|
| 19 |
+
"<extra_id_115>": 374,
|
| 20 |
+
"<extra_id_116>": 375,
|
| 21 |
+
"<extra_id_117>": 376,
|
| 22 |
+
"<extra_id_118>": 377,
|
| 23 |
+
"<extra_id_119>": 378,
|
| 24 |
+
"<extra_id_11>": 270,
|
| 25 |
+
"<extra_id_120>": 379,
|
| 26 |
+
"<extra_id_121>": 380,
|
| 27 |
+
"<extra_id_122>": 381,
|
| 28 |
+
"<extra_id_123>": 382,
|
| 29 |
+
"<extra_id_124>": 383,
|
| 30 |
+
"<extra_id_12>": 271,
|
| 31 |
+
"<extra_id_13>": 272,
|
| 32 |
+
"<extra_id_14>": 273,
|
| 33 |
+
"<extra_id_15>": 274,
|
| 34 |
+
"<extra_id_16>": 275,
|
| 35 |
+
"<extra_id_17>": 276,
|
| 36 |
+
"<extra_id_18>": 277,
|
| 37 |
+
"<extra_id_19>": 278,
|
| 38 |
+
"<extra_id_1>": 260,
|
| 39 |
+
"<extra_id_20>": 279,
|
| 40 |
+
"<extra_id_21>": 280,
|
| 41 |
+
"<extra_id_22>": 281,
|
| 42 |
+
"<extra_id_23>": 282,
|
| 43 |
+
"<extra_id_24>": 283,
|
| 44 |
+
"<extra_id_25>": 284,
|
| 45 |
+
"<extra_id_26>": 285,
|
| 46 |
+
"<extra_id_27>": 286,
|
| 47 |
+
"<extra_id_28>": 287,
|
| 48 |
+
"<extra_id_29>": 288,
|
| 49 |
+
"<extra_id_2>": 261,
|
| 50 |
+
"<extra_id_30>": 289,
|
| 51 |
+
"<extra_id_31>": 290,
|
| 52 |
+
"<extra_id_32>": 291,
|
| 53 |
+
"<extra_id_33>": 292,
|
| 54 |
+
"<extra_id_34>": 293,
|
| 55 |
+
"<extra_id_35>": 294,
|
| 56 |
+
"<extra_id_36>": 295,
|
| 57 |
+
"<extra_id_37>": 296,
|
| 58 |
+
"<extra_id_38>": 297,
|
| 59 |
+
"<extra_id_39>": 298,
|
| 60 |
+
"<extra_id_3>": 262,
|
| 61 |
+
"<extra_id_40>": 299,
|
| 62 |
+
"<extra_id_41>": 300,
|
| 63 |
+
"<extra_id_42>": 301,
|
| 64 |
+
"<extra_id_43>": 302,
|
| 65 |
+
"<extra_id_44>": 303,
|
| 66 |
+
"<extra_id_45>": 304,
|
| 67 |
+
"<extra_id_46>": 305,
|
| 68 |
+
"<extra_id_47>": 306,
|
| 69 |
+
"<extra_id_48>": 307,
|
| 70 |
+
"<extra_id_49>": 308,
|
| 71 |
+
"<extra_id_4>": 263,
|
| 72 |
+
"<extra_id_50>": 309,
|
| 73 |
+
"<extra_id_51>": 310,
|
| 74 |
+
"<extra_id_52>": 311,
|
| 75 |
+
"<extra_id_53>": 312,
|
| 76 |
+
"<extra_id_54>": 313,
|
| 77 |
+
"<extra_id_55>": 314,
|
| 78 |
+
"<extra_id_56>": 315,
|
| 79 |
+
"<extra_id_57>": 316,
|
| 80 |
+
"<extra_id_58>": 317,
|
| 81 |
+
"<extra_id_59>": 318,
|
| 82 |
+
"<extra_id_5>": 264,
|
| 83 |
+
"<extra_id_60>": 319,
|
| 84 |
+
"<extra_id_61>": 320,
|
| 85 |
+
"<extra_id_62>": 321,
|
| 86 |
+
"<extra_id_63>": 322,
|
| 87 |
+
"<extra_id_64>": 323,
|
| 88 |
+
"<extra_id_65>": 324,
|
| 89 |
+
"<extra_id_66>": 325,
|
| 90 |
+
"<extra_id_67>": 326,
|
| 91 |
+
"<extra_id_68>": 327,
|
| 92 |
+
"<extra_id_69>": 328,
|
| 93 |
+
"<extra_id_6>": 265,
|
| 94 |
+
"<extra_id_70>": 329,
|
| 95 |
+
"<extra_id_71>": 330,
|
| 96 |
+
"<extra_id_72>": 331,
|
| 97 |
+
"<extra_id_73>": 332,
|
| 98 |
+
"<extra_id_74>": 333,
|
| 99 |
+
"<extra_id_75>": 334,
|
| 100 |
+
"<extra_id_76>": 335,
|
| 101 |
+
"<extra_id_77>": 336,
|
| 102 |
+
"<extra_id_78>": 337,
|
| 103 |
+
"<extra_id_79>": 338,
|
| 104 |
+
"<extra_id_7>": 266,
|
| 105 |
+
"<extra_id_80>": 339,
|
| 106 |
+
"<extra_id_81>": 340,
|
| 107 |
+
"<extra_id_82>": 341,
|
| 108 |
+
"<extra_id_83>": 342,
|
| 109 |
+
"<extra_id_84>": 343,
|
| 110 |
+
"<extra_id_85>": 344,
|
| 111 |
+
"<extra_id_86>": 345,
|
| 112 |
+
"<extra_id_87>": 346,
|
| 113 |
+
"<extra_id_88>": 347,
|
| 114 |
+
"<extra_id_89>": 348,
|
| 115 |
+
"<extra_id_8>": 267,
|
| 116 |
+
"<extra_id_90>": 349,
|
| 117 |
+
"<extra_id_91>": 350,
|
| 118 |
+
"<extra_id_92>": 351,
|
| 119 |
+
"<extra_id_93>": 352,
|
| 120 |
+
"<extra_id_94>": 353,
|
| 121 |
+
"<extra_id_95>": 354,
|
| 122 |
+
"<extra_id_96>": 355,
|
| 123 |
+
"<extra_id_97>": 356,
|
| 124 |
+
"<extra_id_98>": 357,
|
| 125 |
+
"<extra_id_99>": 358,
|
| 126 |
+
"<extra_id_9>": 268
|
| 127 |
+
}
|
tokenizer/tokenizer_config.json
ADDED
|
@@ -0,0 +1,1292 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "<pad>",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": true,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "</s>",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": true,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "<unk>",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": true,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"259": {
|
| 28 |
+
"content": "<extra_id_0>",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"260": {
|
| 36 |
+
"content": "<extra_id_1>",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
},
|
| 43 |
+
"261": {
|
| 44 |
+
"content": "<extra_id_2>",
|
| 45 |
+
"lstrip": false,
|
| 46 |
+
"normalized": false,
|
| 47 |
+
"rstrip": false,
|
| 48 |
+
"single_word": false,
|
| 49 |
+
"special": true
|
| 50 |
+
},
|
| 51 |
+
"262": {
|
| 52 |
+
"content": "<extra_id_3>",
|
| 53 |
+
"lstrip": false,
|
| 54 |
+
"normalized": false,
|
| 55 |
+
"rstrip": false,
|
| 56 |
+
"single_word": false,
|
| 57 |
+
"special": true
|
| 58 |
+
},
|
| 59 |
+
"263": {
|
| 60 |
+
"content": "<extra_id_4>",
|
| 61 |
+
"lstrip": false,
|
| 62 |
+
"normalized": false,
|
| 63 |
+
"rstrip": false,
|
| 64 |
+
"single_word": false,
|
| 65 |
+
"special": true
|
| 66 |
+
},
|
| 67 |
+
"264": {
|
| 68 |
+
"content": "<extra_id_5>",
|
| 69 |
+
"lstrip": false,
|
| 70 |
+
"normalized": false,
|
| 71 |
+
"rstrip": false,
|
| 72 |
+
"single_word": false,
|
| 73 |
+
"special": true
|
| 74 |
+
},
|
| 75 |
+
"265": {
|
| 76 |
+
"content": "<extra_id_6>",
|
| 77 |
+
"lstrip": false,
|
| 78 |
+
"normalized": false,
|
| 79 |
+
"rstrip": false,
|
| 80 |
+
"single_word": false,
|
| 81 |
+
"special": true
|
| 82 |
+
},
|
| 83 |
+
"266": {
|
| 84 |
+
"content": "<extra_id_7>",
|
| 85 |
+
"lstrip": false,
|
| 86 |
+
"normalized": false,
|
| 87 |
+
"rstrip": false,
|
| 88 |
+
"single_word": false,
|
| 89 |
+
"special": true
|
| 90 |
+
},
|
| 91 |
+
"267": {
|
| 92 |
+
"content": "<extra_id_8>",
|
| 93 |
+
"lstrip": false,
|
| 94 |
+
"normalized": false,
|
| 95 |
+
"rstrip": false,
|
| 96 |
+
"single_word": false,
|
| 97 |
+
"special": true
|
| 98 |
+
},
|
| 99 |
+
"268": {
|
| 100 |
+
"content": "<extra_id_9>",
|
| 101 |
+
"lstrip": false,
|
| 102 |
+
"normalized": false,
|
| 103 |
+
"rstrip": false,
|
| 104 |
+
"single_word": false,
|
| 105 |
+
"special": true
|
| 106 |
+
},
|
| 107 |
+
"269": {
|
| 108 |
+
"content": "<extra_id_10>",
|
| 109 |
+
"lstrip": false,
|
| 110 |
+
"normalized": false,
|
| 111 |
+
"rstrip": false,
|
| 112 |
+
"single_word": false,
|
| 113 |
+
"special": true
|
| 114 |
+
},
|
| 115 |
+
"270": {
|
| 116 |
+
"content": "<extra_id_11>",
|
| 117 |
+
"lstrip": false,
|
| 118 |
+
"normalized": false,
|
| 119 |
+
"rstrip": false,
|
| 120 |
+
"single_word": false,
|
| 121 |
+
"special": true
|
| 122 |
+
},
|
| 123 |
+
"271": {
|
| 124 |
+
"content": "<extra_id_12>",
|
| 125 |
+
"lstrip": false,
|
| 126 |
+
"normalized": false,
|
| 127 |
+
"rstrip": false,
|
| 128 |
+
"single_word": false,
|
| 129 |
+
"special": true
|
| 130 |
+
},
|
| 131 |
+
"272": {
|
| 132 |
+
"content": "<extra_id_13>",
|
| 133 |
+
"lstrip": false,
|
| 134 |
+
"normalized": false,
|
| 135 |
+
"rstrip": false,
|
| 136 |
+
"single_word": false,
|
| 137 |
+
"special": true
|
| 138 |
+
},
|
| 139 |
+
"273": {
|
| 140 |
+
"content": "<extra_id_14>",
|
| 141 |
+
"lstrip": false,
|
| 142 |
+
"normalized": false,
|
| 143 |
+
"rstrip": false,
|
| 144 |
+
"single_word": false,
|
| 145 |
+
"special": true
|
| 146 |
+
},
|
| 147 |
+
"274": {
|
| 148 |
+
"content": "<extra_id_15>",
|
| 149 |
+
"lstrip": false,
|
| 150 |
+
"normalized": false,
|
| 151 |
+
"rstrip": false,
|
| 152 |
+
"single_word": false,
|
| 153 |
+
"special": true
|
| 154 |
+
},
|
| 155 |
+
"275": {
|
| 156 |
+
"content": "<extra_id_16>",
|
| 157 |
+
"lstrip": false,
|
| 158 |
+
"normalized": false,
|
| 159 |
+
"rstrip": false,
|
| 160 |
+
"single_word": false,
|
| 161 |
+
"special": true
|
| 162 |
+
},
|
| 163 |
+
"276": {
|
| 164 |
+
"content": "<extra_id_17>",
|
| 165 |
+
"lstrip": false,
|
| 166 |
+
"normalized": false,
|
| 167 |
+
"rstrip": false,
|
| 168 |
+
"single_word": false,
|
| 169 |
+
"special": true
|
| 170 |
+
},
|
| 171 |
+
"277": {
|
| 172 |
+
"content": "<extra_id_18>",
|
| 173 |
+
"lstrip": false,
|
| 174 |
+
"normalized": false,
|
| 175 |
+
"rstrip": false,
|
| 176 |
+
"single_word": false,
|
| 177 |
+
"special": true
|
| 178 |
+
},
|
| 179 |
+
"278": {
|
| 180 |
+
"content": "<extra_id_19>",
|
| 181 |
+
"lstrip": false,
|
| 182 |
+
"normalized": false,
|
| 183 |
+
"rstrip": false,
|
| 184 |
+
"single_word": false,
|
| 185 |
+
"special": true
|
| 186 |
+
},
|
| 187 |
+
"279": {
|
| 188 |
+
"content": "<extra_id_20>",
|
| 189 |
+
"lstrip": false,
|
| 190 |
+
"normalized": false,
|
| 191 |
+
"rstrip": false,
|
| 192 |
+
"single_word": false,
|
| 193 |
+
"special": true
|
| 194 |
+
},
|
| 195 |
+
"280": {
|
| 196 |
+
"content": "<extra_id_21>",
|
| 197 |
+
"lstrip": false,
|
| 198 |
+
"normalized": false,
|
| 199 |
+
"rstrip": false,
|
| 200 |
+
"single_word": false,
|
| 201 |
+
"special": true
|
| 202 |
+
},
|
| 203 |
+
"281": {
|
| 204 |
+
"content": "<extra_id_22>",
|
| 205 |
+
"lstrip": false,
|
| 206 |
+
"normalized": false,
|
| 207 |
+
"rstrip": false,
|
| 208 |
+
"single_word": false,
|
| 209 |
+
"special": true
|
| 210 |
+
},
|
| 211 |
+
"282": {
|
| 212 |
+
"content": "<extra_id_23>",
|
| 213 |
+
"lstrip": false,
|
| 214 |
+
"normalized": false,
|
| 215 |
+
"rstrip": false,
|
| 216 |
+
"single_word": false,
|
| 217 |
+
"special": true
|
| 218 |
+
},
|
| 219 |
+
"283": {
|
| 220 |
+
"content": "<extra_id_24>",
|
| 221 |
+
"lstrip": false,
|
| 222 |
+
"normalized": false,
|
| 223 |
+
"rstrip": false,
|
| 224 |
+
"single_word": false,
|
| 225 |
+
"special": true
|
| 226 |
+
},
|
| 227 |
+
"284": {
|
| 228 |
+
"content": "<extra_id_25>",
|
| 229 |
+
"lstrip": false,
|
| 230 |
+
"normalized": false,
|
| 231 |
+
"rstrip": false,
|
| 232 |
+
"single_word": false,
|
| 233 |
+
"special": true
|
| 234 |
+
},
|
| 235 |
+
"285": {
|
| 236 |
+
"content": "<extra_id_26>",
|
| 237 |
+
"lstrip": false,
|
| 238 |
+
"normalized": false,
|
| 239 |
+
"rstrip": false,
|
| 240 |
+
"single_word": false,
|
| 241 |
+
"special": true
|
| 242 |
+
},
|
| 243 |
+
"286": {
|
| 244 |
+
"content": "<extra_id_27>",
|
| 245 |
+
"lstrip": false,
|
| 246 |
+
"normalized": false,
|
| 247 |
+
"rstrip": false,
|
| 248 |
+
"single_word": false,
|
| 249 |
+
"special": true
|
| 250 |
+
},
|
| 251 |
+
"287": {
|
| 252 |
+
"content": "<extra_id_28>",
|
| 253 |
+
"lstrip": false,
|
| 254 |
+
"normalized": false,
|
| 255 |
+
"rstrip": false,
|
| 256 |
+
"single_word": false,
|
| 257 |
+
"special": true
|
| 258 |
+
},
|
| 259 |
+
"288": {
|
| 260 |
+
"content": "<extra_id_29>",
|
| 261 |
+
"lstrip": false,
|
| 262 |
+
"normalized": false,
|
| 263 |
+
"rstrip": false,
|
| 264 |
+
"single_word": false,
|
| 265 |
+
"special": true
|
| 266 |
+
},
|
| 267 |
+
"289": {
|
| 268 |
+
"content": "<extra_id_30>",
|
| 269 |
+
"lstrip": false,
|
| 270 |
+
"normalized": false,
|
| 271 |
+
"rstrip": false,
|
| 272 |
+
"single_word": false,
|
| 273 |
+
"special": true
|
| 274 |
+
},
|
| 275 |
+
"290": {
|
| 276 |
+
"content": "<extra_id_31>",
|
| 277 |
+
"lstrip": false,
|
| 278 |
+
"normalized": false,
|
| 279 |
+
"rstrip": false,
|
| 280 |
+
"single_word": false,
|
| 281 |
+
"special": true
|
| 282 |
+
},
|
| 283 |
+
"291": {
|
| 284 |
+
"content": "<extra_id_32>",
|
| 285 |
+
"lstrip": false,
|
| 286 |
+
"normalized": false,
|
| 287 |
+
"rstrip": false,
|
| 288 |
+
"single_word": false,
|
| 289 |
+
"special": true
|
| 290 |
+
},
|
| 291 |
+
"292": {
|
| 292 |
+
"content": "<extra_id_33>",
|
| 293 |
+
"lstrip": false,
|
| 294 |
+
"normalized": false,
|
| 295 |
+
"rstrip": false,
|
| 296 |
+
"single_word": false,
|
| 297 |
+
"special": true
|
| 298 |
+
},
|
| 299 |
+
"293": {
|
| 300 |
+
"content": "<extra_id_34>",
|
| 301 |
+
"lstrip": false,
|
| 302 |
+
"normalized": false,
|
| 303 |
+
"rstrip": false,
|
| 304 |
+
"single_word": false,
|
| 305 |
+
"special": true
|
| 306 |
+
},
|
| 307 |
+
"294": {
|
| 308 |
+
"content": "<extra_id_35>",
|
| 309 |
+
"lstrip": false,
|
| 310 |
+
"normalized": false,
|
| 311 |
+
"rstrip": false,
|
| 312 |
+
"single_word": false,
|
| 313 |
+
"special": true
|
| 314 |
+
},
|
| 315 |
+
"295": {
|
| 316 |
+
"content": "<extra_id_36>",
|
| 317 |
+
"lstrip": false,
|
| 318 |
+
"normalized": false,
|
| 319 |
+
"rstrip": false,
|
| 320 |
+
"single_word": false,
|
| 321 |
+
"special": true
|
| 322 |
+
},
|
| 323 |
+
"296": {
|
| 324 |
+
"content": "<extra_id_37>",
|
| 325 |
+
"lstrip": false,
|
| 326 |
+
"normalized": false,
|
| 327 |
+
"rstrip": false,
|
| 328 |
+
"single_word": false,
|
| 329 |
+
"special": true
|
| 330 |
+
},
|
| 331 |
+
"297": {
|
| 332 |
+
"content": "<extra_id_38>",
|
| 333 |
+
"lstrip": false,
|
| 334 |
+
"normalized": false,
|
| 335 |
+
"rstrip": false,
|
| 336 |
+
"single_word": false,
|
| 337 |
+
"special": true
|
| 338 |
+
},
|
| 339 |
+
"298": {
|
| 340 |
+
"content": "<extra_id_39>",
|
| 341 |
+
"lstrip": false,
|
| 342 |
+
"normalized": false,
|
| 343 |
+
"rstrip": false,
|
| 344 |
+
"single_word": false,
|
| 345 |
+
"special": true
|
| 346 |
+
},
|
| 347 |
+
"299": {
|
| 348 |
+
"content": "<extra_id_40>",
|
| 349 |
+
"lstrip": false,
|
| 350 |
+
"normalized": false,
|
| 351 |
+
"rstrip": false,
|
| 352 |
+
"single_word": false,
|
| 353 |
+
"special": true
|
| 354 |
+
},
|
| 355 |
+
"300": {
|
| 356 |
+
"content": "<extra_id_41>",
|
| 357 |
+
"lstrip": false,
|
| 358 |
+
"normalized": false,
|
| 359 |
+
"rstrip": false,
|
| 360 |
+
"single_word": false,
|
| 361 |
+
"special": true
|
| 362 |
+
},
|
| 363 |
+
"301": {
|
| 364 |
+
"content": "<extra_id_42>",
|
| 365 |
+
"lstrip": false,
|
| 366 |
+
"normalized": false,
|
| 367 |
+
"rstrip": false,
|
| 368 |
+
"single_word": false,
|
| 369 |
+
"special": true
|
| 370 |
+
},
|
| 371 |
+
"302": {
|
| 372 |
+
"content": "<extra_id_43>",
|
| 373 |
+
"lstrip": false,
|
| 374 |
+
"normalized": false,
|
| 375 |
+
"rstrip": false,
|
| 376 |
+
"single_word": false,
|
| 377 |
+
"special": true
|
| 378 |
+
},
|
| 379 |
+
"303": {
|
| 380 |
+
"content": "<extra_id_44>",
|
| 381 |
+
"lstrip": false,
|
| 382 |
+
"normalized": false,
|
| 383 |
+
"rstrip": false,
|
| 384 |
+
"single_word": false,
|
| 385 |
+
"special": true
|
| 386 |
+
},
|
| 387 |
+
"304": {
|
| 388 |
+
"content": "<extra_id_45>",
|
| 389 |
+
"lstrip": false,
|
| 390 |
+
"normalized": false,
|
| 391 |
+
"rstrip": false,
|
| 392 |
+
"single_word": false,
|
| 393 |
+
"special": true
|
| 394 |
+
},
|
| 395 |
+
"305": {
|
| 396 |
+
"content": "<extra_id_46>",
|
| 397 |
+
"lstrip": false,
|
| 398 |
+
"normalized": false,
|
| 399 |
+
"rstrip": false,
|
| 400 |
+
"single_word": false,
|
| 401 |
+
"special": true
|
| 402 |
+
},
|
| 403 |
+
"306": {
|
| 404 |
+
"content": "<extra_id_47>",
|
| 405 |
+
"lstrip": false,
|
| 406 |
+
"normalized": false,
|
| 407 |
+
"rstrip": false,
|
| 408 |
+
"single_word": false,
|
| 409 |
+
"special": true
|
| 410 |
+
},
|
| 411 |
+
"307": {
|
| 412 |
+
"content": "<extra_id_48>",
|
| 413 |
+
"lstrip": false,
|
| 414 |
+
"normalized": false,
|
| 415 |
+
"rstrip": false,
|
| 416 |
+
"single_word": false,
|
| 417 |
+
"special": true
|
| 418 |
+
},
|
| 419 |
+
"308": {
|
| 420 |
+
"content": "<extra_id_49>",
|
| 421 |
+
"lstrip": false,
|
| 422 |
+
"normalized": false,
|
| 423 |
+
"rstrip": false,
|
| 424 |
+
"single_word": false,
|
| 425 |
+
"special": true
|
| 426 |
+
},
|
| 427 |
+
"309": {
|
| 428 |
+
"content": "<extra_id_50>",
|
| 429 |
+
"lstrip": false,
|
| 430 |
+
"normalized": false,
|
| 431 |
+
"rstrip": false,
|
| 432 |
+
"single_word": false,
|
| 433 |
+
"special": true
|
| 434 |
+
},
|
| 435 |
+
"310": {
|
| 436 |
+
"content": "<extra_id_51>",
|
| 437 |
+
"lstrip": false,
|
| 438 |
+
"normalized": false,
|
| 439 |
+
"rstrip": false,
|
| 440 |
+
"single_word": false,
|
| 441 |
+
"special": true
|
| 442 |
+
},
|
| 443 |
+
"311": {
|
| 444 |
+
"content": "<extra_id_52>",
|
| 445 |
+
"lstrip": false,
|
| 446 |
+
"normalized": false,
|
| 447 |
+
"rstrip": false,
|
| 448 |
+
"single_word": false,
|
| 449 |
+
"special": true
|
| 450 |
+
},
|
| 451 |
+
"312": {
|
| 452 |
+
"content": "<extra_id_53>",
|
| 453 |
+
"lstrip": false,
|
| 454 |
+
"normalized": false,
|
| 455 |
+
"rstrip": false,
|
| 456 |
+
"single_word": false,
|
| 457 |
+
"special": true
|
| 458 |
+
},
|
| 459 |
+
"313": {
|
| 460 |
+
"content": "<extra_id_54>",
|
| 461 |
+
"lstrip": false,
|
| 462 |
+
"normalized": false,
|
| 463 |
+
"rstrip": false,
|
| 464 |
+
"single_word": false,
|
| 465 |
+
"special": true
|
| 466 |
+
},
|
| 467 |
+
"314": {
|
| 468 |
+
"content": "<extra_id_55>",
|
| 469 |
+
"lstrip": false,
|
| 470 |
+
"normalized": false,
|
| 471 |
+
"rstrip": false,
|
| 472 |
+
"single_word": false,
|
| 473 |
+
"special": true
|
| 474 |
+
},
|
| 475 |
+
"315": {
|
| 476 |
+
"content": "<extra_id_56>",
|
| 477 |
+
"lstrip": false,
|
| 478 |
+
"normalized": false,
|
| 479 |
+
"rstrip": false,
|
| 480 |
+
"single_word": false,
|
| 481 |
+
"special": true
|
| 482 |
+
},
|
| 483 |
+
"316": {
|
| 484 |
+
"content": "<extra_id_57>",
|
| 485 |
+
"lstrip": false,
|
| 486 |
+
"normalized": false,
|
| 487 |
+
"rstrip": false,
|
| 488 |
+
"single_word": false,
|
| 489 |
+
"special": true
|
| 490 |
+
},
|
| 491 |
+
"317": {
|
| 492 |
+
"content": "<extra_id_58>",
|
| 493 |
+
"lstrip": false,
|
| 494 |
+
"normalized": false,
|
| 495 |
+
"rstrip": false,
|
| 496 |
+
"single_word": false,
|
| 497 |
+
"special": true
|
| 498 |
+
},
|
| 499 |
+
"318": {
|
| 500 |
+
"content": "<extra_id_59>",
|
| 501 |
+
"lstrip": false,
|
| 502 |
+
"normalized": false,
|
| 503 |
+
"rstrip": false,
|
| 504 |
+
"single_word": false,
|
| 505 |
+
"special": true
|
| 506 |
+
},
|
| 507 |
+
"319": {
|
| 508 |
+
"content": "<extra_id_60>",
|
| 509 |
+
"lstrip": false,
|
| 510 |
+
"normalized": false,
|
| 511 |
+
"rstrip": false,
|
| 512 |
+
"single_word": false,
|
| 513 |
+
"special": true
|
| 514 |
+
},
|
| 515 |
+
"320": {
|
| 516 |
+
"content": "<extra_id_61>",
|
| 517 |
+
"lstrip": false,
|
| 518 |
+
"normalized": false,
|
| 519 |
+
"rstrip": false,
|
| 520 |
+
"single_word": false,
|
| 521 |
+
"special": true
|
| 522 |
+
},
|
| 523 |
+
"321": {
|
| 524 |
+
"content": "<extra_id_62>",
|
| 525 |
+
"lstrip": false,
|
| 526 |
+
"normalized": false,
|
| 527 |
+
"rstrip": false,
|
| 528 |
+
"single_word": false,
|
| 529 |
+
"special": true
|
| 530 |
+
},
|
| 531 |
+
"322": {
|
| 532 |
+
"content": "<extra_id_63>",
|
| 533 |
+
"lstrip": false,
|
| 534 |
+
"normalized": false,
|
| 535 |
+
"rstrip": false,
|
| 536 |
+
"single_word": false,
|
| 537 |
+
"special": true
|
| 538 |
+
},
|
| 539 |
+
"323": {
|
| 540 |
+
"content": "<extra_id_64>",
|
| 541 |
+
"lstrip": false,
|
| 542 |
+
"normalized": false,
|
| 543 |
+
"rstrip": false,
|
| 544 |
+
"single_word": false,
|
| 545 |
+
"special": true
|
| 546 |
+
},
|
| 547 |
+
"324": {
|
| 548 |
+
"content": "<extra_id_65>",
|
| 549 |
+
"lstrip": false,
|
| 550 |
+
"normalized": false,
|
| 551 |
+
"rstrip": false,
|
| 552 |
+
"single_word": false,
|
| 553 |
+
"special": true
|
| 554 |
+
},
|
| 555 |
+
"325": {
|
| 556 |
+
"content": "<extra_id_66>",
|
| 557 |
+
"lstrip": false,
|
| 558 |
+
"normalized": false,
|
| 559 |
+
"rstrip": false,
|
| 560 |
+
"single_word": false,
|
| 561 |
+
"special": true
|
| 562 |
+
},
|
| 563 |
+
"326": {
|
| 564 |
+
"content": "<extra_id_67>",
|
| 565 |
+
"lstrip": false,
|
| 566 |
+
"normalized": false,
|
| 567 |
+
"rstrip": false,
|
| 568 |
+
"single_word": false,
|
| 569 |
+
"special": true
|
| 570 |
+
},
|
| 571 |
+
"327": {
|
| 572 |
+
"content": "<extra_id_68>",
|
| 573 |
+
"lstrip": false,
|
| 574 |
+
"normalized": false,
|
| 575 |
+
"rstrip": false,
|
| 576 |
+
"single_word": false,
|
| 577 |
+
"special": true
|
| 578 |
+
},
|
| 579 |
+
"328": {
|
| 580 |
+
"content": "<extra_id_69>",
|
| 581 |
+
"lstrip": false,
|
| 582 |
+
"normalized": false,
|
| 583 |
+
"rstrip": false,
|
| 584 |
+
"single_word": false,
|
| 585 |
+
"special": true
|
| 586 |
+
},
|
| 587 |
+
"329": {
|
| 588 |
+
"content": "<extra_id_70>",
|
| 589 |
+
"lstrip": false,
|
| 590 |
+
"normalized": false,
|
| 591 |
+
"rstrip": false,
|
| 592 |
+
"single_word": false,
|
| 593 |
+
"special": true
|
| 594 |
+
},
|
| 595 |
+
"330": {
|
| 596 |
+
"content": "<extra_id_71>",
|
| 597 |
+
"lstrip": false,
|
| 598 |
+
"normalized": false,
|
| 599 |
+
"rstrip": false,
|
| 600 |
+
"single_word": false,
|
| 601 |
+
"special": true
|
| 602 |
+
},
|
| 603 |
+
"331": {
|
| 604 |
+
"content": "<extra_id_72>",
|
| 605 |
+
"lstrip": false,
|
| 606 |
+
"normalized": false,
|
| 607 |
+
"rstrip": false,
|
| 608 |
+
"single_word": false,
|
| 609 |
+
"special": true
|
| 610 |
+
},
|
| 611 |
+
"332": {
|
| 612 |
+
"content": "<extra_id_73>",
|
| 613 |
+
"lstrip": false,
|
| 614 |
+
"normalized": false,
|
| 615 |
+
"rstrip": false,
|
| 616 |
+
"single_word": false,
|
| 617 |
+
"special": true
|
| 618 |
+
},
|
| 619 |
+
"333": {
|
| 620 |
+
"content": "<extra_id_74>",
|
| 621 |
+
"lstrip": false,
|
| 622 |
+
"normalized": false,
|
| 623 |
+
"rstrip": false,
|
| 624 |
+
"single_word": false,
|
| 625 |
+
"special": true
|
| 626 |
+
},
|
| 627 |
+
"334": {
|
| 628 |
+
"content": "<extra_id_75>",
|
| 629 |
+
"lstrip": false,
|
| 630 |
+
"normalized": false,
|
| 631 |
+
"rstrip": false,
|
| 632 |
+
"single_word": false,
|
| 633 |
+
"special": true
|
| 634 |
+
},
|
| 635 |
+
"335": {
|
| 636 |
+
"content": "<extra_id_76>",
|
| 637 |
+
"lstrip": false,
|
| 638 |
+
"normalized": false,
|
| 639 |
+
"rstrip": false,
|
| 640 |
+
"single_word": false,
|
| 641 |
+
"special": true
|
| 642 |
+
},
|
| 643 |
+
"336": {
|
| 644 |
+
"content": "<extra_id_77>",
|
| 645 |
+
"lstrip": false,
|
| 646 |
+
"normalized": false,
|
| 647 |
+
"rstrip": false,
|
| 648 |
+
"single_word": false,
|
| 649 |
+
"special": true
|
| 650 |
+
},
|
| 651 |
+
"337": {
|
| 652 |
+
"content": "<extra_id_78>",
|
| 653 |
+
"lstrip": false,
|
| 654 |
+
"normalized": false,
|
| 655 |
+
"rstrip": false,
|
| 656 |
+
"single_word": false,
|
| 657 |
+
"special": true
|
| 658 |
+
},
|
| 659 |
+
"338": {
|
| 660 |
+
"content": "<extra_id_79>",
|
| 661 |
+
"lstrip": false,
|
| 662 |
+
"normalized": false,
|
| 663 |
+
"rstrip": false,
|
| 664 |
+
"single_word": false,
|
| 665 |
+
"special": true
|
| 666 |
+
},
|
| 667 |
+
"339": {
|
| 668 |
+
"content": "<extra_id_80>",
|
| 669 |
+
"lstrip": false,
|
| 670 |
+
"normalized": false,
|
| 671 |
+
"rstrip": false,
|
| 672 |
+
"single_word": false,
|
| 673 |
+
"special": true
|
| 674 |
+
},
|
| 675 |
+
"340": {
|
| 676 |
+
"content": "<extra_id_81>",
|
| 677 |
+
"lstrip": false,
|
| 678 |
+
"normalized": false,
|
| 679 |
+
"rstrip": false,
|
| 680 |
+
"single_word": false,
|
| 681 |
+
"special": true
|
| 682 |
+
},
|
| 683 |
+
"341": {
|
| 684 |
+
"content": "<extra_id_82>",
|
| 685 |
+
"lstrip": false,
|
| 686 |
+
"normalized": false,
|
| 687 |
+
"rstrip": false,
|
| 688 |
+
"single_word": false,
|
| 689 |
+
"special": true
|
| 690 |
+
},
|
| 691 |
+
"342": {
|
| 692 |
+
"content": "<extra_id_83>",
|
| 693 |
+
"lstrip": false,
|
| 694 |
+
"normalized": false,
|
| 695 |
+
"rstrip": false,
|
| 696 |
+
"single_word": false,
|
| 697 |
+
"special": true
|
| 698 |
+
},
|
| 699 |
+
"343": {
|
| 700 |
+
"content": "<extra_id_84>",
|
| 701 |
+
"lstrip": false,
|
| 702 |
+
"normalized": false,
|
| 703 |
+
"rstrip": false,
|
| 704 |
+
"single_word": false,
|
| 705 |
+
"special": true
|
| 706 |
+
},
|
| 707 |
+
"344": {
|
| 708 |
+
"content": "<extra_id_85>",
|
| 709 |
+
"lstrip": false,
|
| 710 |
+
"normalized": false,
|
| 711 |
+
"rstrip": false,
|
| 712 |
+
"single_word": false,
|
| 713 |
+
"special": true
|
| 714 |
+
},
|
| 715 |
+
"345": {
|
| 716 |
+
"content": "<extra_id_86>",
|
| 717 |
+
"lstrip": false,
|
| 718 |
+
"normalized": false,
|
| 719 |
+
"rstrip": false,
|
| 720 |
+
"single_word": false,
|
| 721 |
+
"special": true
|
| 722 |
+
},
|
| 723 |
+
"346": {
|
| 724 |
+
"content": "<extra_id_87>",
|
| 725 |
+
"lstrip": false,
|
| 726 |
+
"normalized": false,
|
| 727 |
+
"rstrip": false,
|
| 728 |
+
"single_word": false,
|
| 729 |
+
"special": true
|
| 730 |
+
},
|
| 731 |
+
"347": {
|
| 732 |
+
"content": "<extra_id_88>",
|
| 733 |
+
"lstrip": false,
|
| 734 |
+
"normalized": false,
|
| 735 |
+
"rstrip": false,
|
| 736 |
+
"single_word": false,
|
| 737 |
+
"special": true
|
| 738 |
+
},
|
| 739 |
+
"348": {
|
| 740 |
+
"content": "<extra_id_89>",
|
| 741 |
+
"lstrip": false,
|
| 742 |
+
"normalized": false,
|
| 743 |
+
"rstrip": false,
|
| 744 |
+
"single_word": false,
|
| 745 |
+
"special": true
|
| 746 |
+
},
|
| 747 |
+
"349": {
|
| 748 |
+
"content": "<extra_id_90>",
|
| 749 |
+
"lstrip": false,
|
| 750 |
+
"normalized": false,
|
| 751 |
+
"rstrip": false,
|
| 752 |
+
"single_word": false,
|
| 753 |
+
"special": true
|
| 754 |
+
},
|
| 755 |
+
"350": {
|
| 756 |
+
"content": "<extra_id_91>",
|
| 757 |
+
"lstrip": false,
|
| 758 |
+
"normalized": false,
|
| 759 |
+
"rstrip": false,
|
| 760 |
+
"single_word": false,
|
| 761 |
+
"special": true
|
| 762 |
+
},
|
| 763 |
+
"351": {
|
| 764 |
+
"content": "<extra_id_92>",
|
| 765 |
+
"lstrip": false,
|
| 766 |
+
"normalized": false,
|
| 767 |
+
"rstrip": false,
|
| 768 |
+
"single_word": false,
|
| 769 |
+
"special": true
|
| 770 |
+
},
|
| 771 |
+
"352": {
|
| 772 |
+
"content": "<extra_id_93>",
|
| 773 |
+
"lstrip": false,
|
| 774 |
+
"normalized": false,
|
| 775 |
+
"rstrip": false,
|
| 776 |
+
"single_word": false,
|
| 777 |
+
"special": true
|
| 778 |
+
},
|
| 779 |
+
"353": {
|
| 780 |
+
"content": "<extra_id_94>",
|
| 781 |
+
"lstrip": false,
|
| 782 |
+
"normalized": false,
|
| 783 |
+
"rstrip": false,
|
| 784 |
+
"single_word": false,
|
| 785 |
+
"special": true
|
| 786 |
+
},
|
| 787 |
+
"354": {
|
| 788 |
+
"content": "<extra_id_95>",
|
| 789 |
+
"lstrip": false,
|
| 790 |
+
"normalized": false,
|
| 791 |
+
"rstrip": false,
|
| 792 |
+
"single_word": false,
|
| 793 |
+
"special": true
|
| 794 |
+
},
|
| 795 |
+
"355": {
|
| 796 |
+
"content": "<extra_id_96>",
|
| 797 |
+
"lstrip": false,
|
| 798 |
+
"normalized": false,
|
| 799 |
+
"rstrip": false,
|
| 800 |
+
"single_word": false,
|
| 801 |
+
"special": true
|
| 802 |
+
},
|
| 803 |
+
"356": {
|
| 804 |
+
"content": "<extra_id_97>",
|
| 805 |
+
"lstrip": false,
|
| 806 |
+
"normalized": false,
|
| 807 |
+
"rstrip": false,
|
| 808 |
+
"single_word": false,
|
| 809 |
+
"special": true
|
| 810 |
+
},
|
| 811 |
+
"357": {
|
| 812 |
+
"content": "<extra_id_98>",
|
| 813 |
+
"lstrip": false,
|
| 814 |
+
"normalized": false,
|
| 815 |
+
"rstrip": false,
|
| 816 |
+
"single_word": false,
|
| 817 |
+
"special": true
|
| 818 |
+
},
|
| 819 |
+
"358": {
|
| 820 |
+
"content": "<extra_id_99>",
|
| 821 |
+
"lstrip": false,
|
| 822 |
+
"normalized": false,
|
| 823 |
+
"rstrip": false,
|
| 824 |
+
"single_word": false,
|
| 825 |
+
"special": true
|
| 826 |
+
},
|
| 827 |
+
"359": {
|
| 828 |
+
"content": "<extra_id_100>",
|
| 829 |
+
"lstrip": false,
|
| 830 |
+
"normalized": false,
|
| 831 |
+
"rstrip": false,
|
| 832 |
+
"single_word": false,
|
| 833 |
+
"special": true
|
| 834 |
+
},
|
| 835 |
+
"360": {
|
| 836 |
+
"content": "<extra_id_101>",
|
| 837 |
+
"lstrip": false,
|
| 838 |
+
"normalized": false,
|
| 839 |
+
"rstrip": false,
|
| 840 |
+
"single_word": false,
|
| 841 |
+
"special": true
|
| 842 |
+
},
|
| 843 |
+
"361": {
|
| 844 |
+
"content": "<extra_id_102>",
|
| 845 |
+
"lstrip": false,
|
| 846 |
+
"normalized": false,
|
| 847 |
+
"rstrip": false,
|
| 848 |
+
"single_word": false,
|
| 849 |
+
"special": true
|
| 850 |
+
},
|
| 851 |
+
"362": {
|
| 852 |
+
"content": "<extra_id_103>",
|
| 853 |
+
"lstrip": false,
|
| 854 |
+
"normalized": false,
|
| 855 |
+
"rstrip": false,
|
| 856 |
+
"single_word": false,
|
| 857 |
+
"special": true
|
| 858 |
+
},
|
| 859 |
+
"363": {
|
| 860 |
+
"content": "<extra_id_104>",
|
| 861 |
+
"lstrip": false,
|
| 862 |
+
"normalized": false,
|
| 863 |
+
"rstrip": false,
|
| 864 |
+
"single_word": false,
|
| 865 |
+
"special": true
|
| 866 |
+
},
|
| 867 |
+
"364": {
|
| 868 |
+
"content": "<extra_id_105>",
|
| 869 |
+
"lstrip": false,
|
| 870 |
+
"normalized": false,
|
| 871 |
+
"rstrip": false,
|
| 872 |
+
"single_word": false,
|
| 873 |
+
"special": true
|
| 874 |
+
},
|
| 875 |
+
"365": {
|
| 876 |
+
"content": "<extra_id_106>",
|
| 877 |
+
"lstrip": false,
|
| 878 |
+
"normalized": false,
|
| 879 |
+
"rstrip": false,
|
| 880 |
+
"single_word": false,
|
| 881 |
+
"special": true
|
| 882 |
+
},
|
| 883 |
+
"366": {
|
| 884 |
+
"content": "<extra_id_107>",
|
| 885 |
+
"lstrip": false,
|
| 886 |
+
"normalized": false,
|
| 887 |
+
"rstrip": false,
|
| 888 |
+
"single_word": false,
|
| 889 |
+
"special": true
|
| 890 |
+
},
|
| 891 |
+
"367": {
|
| 892 |
+
"content": "<extra_id_108>",
|
| 893 |
+
"lstrip": false,
|
| 894 |
+
"normalized": false,
|
| 895 |
+
"rstrip": false,
|
| 896 |
+
"single_word": false,
|
| 897 |
+
"special": true
|
| 898 |
+
},
|
| 899 |
+
"368": {
|
| 900 |
+
"content": "<extra_id_109>",
|
| 901 |
+
"lstrip": false,
|
| 902 |
+
"normalized": false,
|
| 903 |
+
"rstrip": false,
|
| 904 |
+
"single_word": false,
|
| 905 |
+
"special": true
|
| 906 |
+
},
|
| 907 |
+
"369": {
|
| 908 |
+
"content": "<extra_id_110>",
|
| 909 |
+
"lstrip": false,
|
| 910 |
+
"normalized": false,
|
| 911 |
+
"rstrip": false,
|
| 912 |
+
"single_word": false,
|
| 913 |
+
"special": true
|
| 914 |
+
},
|
| 915 |
+
"370": {
|
| 916 |
+
"content": "<extra_id_111>",
|
| 917 |
+
"lstrip": false,
|
| 918 |
+
"normalized": false,
|
| 919 |
+
"rstrip": false,
|
| 920 |
+
"single_word": false,
|
| 921 |
+
"special": true
|
| 922 |
+
},
|
| 923 |
+
"371": {
|
| 924 |
+
"content": "<extra_id_112>",
|
| 925 |
+
"lstrip": false,
|
| 926 |
+
"normalized": false,
|
| 927 |
+
"rstrip": false,
|
| 928 |
+
"single_word": false,
|
| 929 |
+
"special": true
|
| 930 |
+
},
|
| 931 |
+
"372": {
|
| 932 |
+
"content": "<extra_id_113>",
|
| 933 |
+
"lstrip": false,
|
| 934 |
+
"normalized": false,
|
| 935 |
+
"rstrip": false,
|
| 936 |
+
"single_word": false,
|
| 937 |
+
"special": true
|
| 938 |
+
},
|
| 939 |
+
"373": {
|
| 940 |
+
"content": "<extra_id_114>",
|
| 941 |
+
"lstrip": false,
|
| 942 |
+
"normalized": false,
|
| 943 |
+
"rstrip": false,
|
| 944 |
+
"single_word": false,
|
| 945 |
+
"special": true
|
| 946 |
+
},
|
| 947 |
+
"374": {
|
| 948 |
+
"content": "<extra_id_115>",
|
| 949 |
+
"lstrip": false,
|
| 950 |
+
"normalized": false,
|
| 951 |
+
"rstrip": false,
|
| 952 |
+
"single_word": false,
|
| 953 |
+
"special": true
|
| 954 |
+
},
|
| 955 |
+
"375": {
|
| 956 |
+
"content": "<extra_id_116>",
|
| 957 |
+
"lstrip": false,
|
| 958 |
+
"normalized": false,
|
| 959 |
+
"rstrip": false,
|
| 960 |
+
"single_word": false,
|
| 961 |
+
"special": true
|
| 962 |
+
},
|
| 963 |
+
"376": {
|
| 964 |
+
"content": "<extra_id_117>",
|
| 965 |
+
"lstrip": false,
|
| 966 |
+
"normalized": false,
|
| 967 |
+
"rstrip": false,
|
| 968 |
+
"single_word": false,
|
| 969 |
+
"special": true
|
| 970 |
+
},
|
| 971 |
+
"377": {
|
| 972 |
+
"content": "<extra_id_118>",
|
| 973 |
+
"lstrip": false,
|
| 974 |
+
"normalized": false,
|
| 975 |
+
"rstrip": false,
|
| 976 |
+
"single_word": false,
|
| 977 |
+
"special": true
|
| 978 |
+
},
|
| 979 |
+
"378": {
|
| 980 |
+
"content": "<extra_id_119>",
|
| 981 |
+
"lstrip": false,
|
| 982 |
+
"normalized": false,
|
| 983 |
+
"rstrip": false,
|
| 984 |
+
"single_word": false,
|
| 985 |
+
"special": true
|
| 986 |
+
},
|
| 987 |
+
"379": {
|
| 988 |
+
"content": "<extra_id_120>",
|
| 989 |
+
"lstrip": false,
|
| 990 |
+
"normalized": false,
|
| 991 |
+
"rstrip": false,
|
| 992 |
+
"single_word": false,
|
| 993 |
+
"special": true
|
| 994 |
+
},
|
| 995 |
+
"380": {
|
| 996 |
+
"content": "<extra_id_121>",
|
| 997 |
+
"lstrip": false,
|
| 998 |
+
"normalized": false,
|
| 999 |
+
"rstrip": false,
|
| 1000 |
+
"single_word": false,
|
| 1001 |
+
"special": true
|
| 1002 |
+
},
|
| 1003 |
+
"381": {
|
| 1004 |
+
"content": "<extra_id_122>",
|
| 1005 |
+
"lstrip": false,
|
| 1006 |
+
"normalized": false,
|
| 1007 |
+
"rstrip": false,
|
| 1008 |
+
"single_word": false,
|
| 1009 |
+
"special": true
|
| 1010 |
+
},
|
| 1011 |
+
"382": {
|
| 1012 |
+
"content": "<extra_id_123>",
|
| 1013 |
+
"lstrip": false,
|
| 1014 |
+
"normalized": false,
|
| 1015 |
+
"rstrip": false,
|
| 1016 |
+
"single_word": false,
|
| 1017 |
+
"special": true
|
| 1018 |
+
},
|
| 1019 |
+
"383": {
|
| 1020 |
+
"content": "<extra_id_124>",
|
| 1021 |
+
"lstrip": false,
|
| 1022 |
+
"normalized": false,
|
| 1023 |
+
"rstrip": false,
|
| 1024 |
+
"single_word": false,
|
| 1025 |
+
"special": true
|
| 1026 |
+
}
|
| 1027 |
+
},
|
| 1028 |
+
"additional_special_tokens": [
|
| 1029 |
+
"<extra_id_0>",
|
| 1030 |
+
"<extra_id_1>",
|
| 1031 |
+
"<extra_id_2>",
|
| 1032 |
+
"<extra_id_3>",
|
| 1033 |
+
"<extra_id_4>",
|
| 1034 |
+
"<extra_id_5>",
|
| 1035 |
+
"<extra_id_6>",
|
| 1036 |
+
"<extra_id_7>",
|
| 1037 |
+
"<extra_id_8>",
|
| 1038 |
+
"<extra_id_9>",
|
| 1039 |
+
"<extra_id_10>",
|
| 1040 |
+
"<extra_id_11>",
|
| 1041 |
+
"<extra_id_12>",
|
| 1042 |
+
"<extra_id_13>",
|
| 1043 |
+
"<extra_id_14>",
|
| 1044 |
+
"<extra_id_15>",
|
| 1045 |
+
"<extra_id_16>",
|
| 1046 |
+
"<extra_id_17>",
|
| 1047 |
+
"<extra_id_18>",
|
| 1048 |
+
"<extra_id_19>",
|
| 1049 |
+
"<extra_id_20>",
|
| 1050 |
+
"<extra_id_21>",
|
| 1051 |
+
"<extra_id_22>",
|
| 1052 |
+
"<extra_id_23>",
|
| 1053 |
+
"<extra_id_24>",
|
| 1054 |
+
"<extra_id_25>",
|
| 1055 |
+
"<extra_id_26>",
|
| 1056 |
+
"<extra_id_27>",
|
| 1057 |
+
"<extra_id_28>",
|
| 1058 |
+
"<extra_id_29>",
|
| 1059 |
+
"<extra_id_30>",
|
| 1060 |
+
"<extra_id_31>",
|
| 1061 |
+
"<extra_id_32>",
|
| 1062 |
+
"<extra_id_33>",
|
| 1063 |
+
"<extra_id_34>",
|
| 1064 |
+
"<extra_id_35>",
|
| 1065 |
+
"<extra_id_36>",
|
| 1066 |
+
"<extra_id_37>",
|
| 1067 |
+
"<extra_id_38>",
|
| 1068 |
+
"<extra_id_39>",
|
| 1069 |
+
"<extra_id_40>",
|
| 1070 |
+
"<extra_id_41>",
|
| 1071 |
+
"<extra_id_42>",
|
| 1072 |
+
"<extra_id_43>",
|
| 1073 |
+
"<extra_id_44>",
|
| 1074 |
+
"<extra_id_45>",
|
| 1075 |
+
"<extra_id_46>",
|
| 1076 |
+
"<extra_id_47>",
|
| 1077 |
+
"<extra_id_48>",
|
| 1078 |
+
"<extra_id_49>",
|
| 1079 |
+
"<extra_id_50>",
|
| 1080 |
+
"<extra_id_51>",
|
| 1081 |
+
"<extra_id_52>",
|
| 1082 |
+
"<extra_id_53>",
|
| 1083 |
+
"<extra_id_54>",
|
| 1084 |
+
"<extra_id_55>",
|
| 1085 |
+
"<extra_id_56>",
|
| 1086 |
+
"<extra_id_57>",
|
| 1087 |
+
"<extra_id_58>",
|
| 1088 |
+
"<extra_id_59>",
|
| 1089 |
+
"<extra_id_60>",
|
| 1090 |
+
"<extra_id_61>",
|
| 1091 |
+
"<extra_id_62>",
|
| 1092 |
+
"<extra_id_63>",
|
| 1093 |
+
"<extra_id_64>",
|
| 1094 |
+
"<extra_id_65>",
|
| 1095 |
+
"<extra_id_66>",
|
| 1096 |
+
"<extra_id_67>",
|
| 1097 |
+
"<extra_id_68>",
|
| 1098 |
+
"<extra_id_69>",
|
| 1099 |
+
"<extra_id_70>",
|
| 1100 |
+
"<extra_id_71>",
|
| 1101 |
+
"<extra_id_72>",
|
| 1102 |
+
"<extra_id_73>",
|
| 1103 |
+
"<extra_id_74>",
|
| 1104 |
+
"<extra_id_75>",
|
| 1105 |
+
"<extra_id_76>",
|
| 1106 |
+
"<extra_id_77>",
|
| 1107 |
+
"<extra_id_78>",
|
| 1108 |
+
"<extra_id_79>",
|
| 1109 |
+
"<extra_id_80>",
|
| 1110 |
+
"<extra_id_81>",
|
| 1111 |
+
"<extra_id_82>",
|
| 1112 |
+
"<extra_id_83>",
|
| 1113 |
+
"<extra_id_84>",
|
| 1114 |
+
"<extra_id_85>",
|
| 1115 |
+
"<extra_id_86>",
|
| 1116 |
+
"<extra_id_87>",
|
| 1117 |
+
"<extra_id_88>",
|
| 1118 |
+
"<extra_id_89>",
|
| 1119 |
+
"<extra_id_90>",
|
| 1120 |
+
"<extra_id_91>",
|
| 1121 |
+
"<extra_id_92>",
|
| 1122 |
+
"<extra_id_93>",
|
| 1123 |
+
"<extra_id_94>",
|
| 1124 |
+
"<extra_id_95>",
|
| 1125 |
+
"<extra_id_96>",
|
| 1126 |
+
"<extra_id_97>",
|
| 1127 |
+
"<extra_id_98>",
|
| 1128 |
+
"<extra_id_99>",
|
| 1129 |
+
"<extra_id_100>",
|
| 1130 |
+
"<extra_id_101>",
|
| 1131 |
+
"<extra_id_102>",
|
| 1132 |
+
"<extra_id_103>",
|
| 1133 |
+
"<extra_id_104>",
|
| 1134 |
+
"<extra_id_105>",
|
| 1135 |
+
"<extra_id_106>",
|
| 1136 |
+
"<extra_id_107>",
|
| 1137 |
+
"<extra_id_108>",
|
| 1138 |
+
"<extra_id_109>",
|
| 1139 |
+
"<extra_id_110>",
|
| 1140 |
+
"<extra_id_111>",
|
| 1141 |
+
"<extra_id_112>",
|
| 1142 |
+
"<extra_id_113>",
|
| 1143 |
+
"<extra_id_114>",
|
| 1144 |
+
"<extra_id_115>",
|
| 1145 |
+
"<extra_id_116>",
|
| 1146 |
+
"<extra_id_117>",
|
| 1147 |
+
"<extra_id_118>",
|
| 1148 |
+
"<extra_id_119>",
|
| 1149 |
+
"<extra_id_120>",
|
| 1150 |
+
"<extra_id_121>",
|
| 1151 |
+
"<extra_id_122>",
|
| 1152 |
+
"<extra_id_123>",
|
| 1153 |
+
"<extra_id_124>"
|
| 1154 |
+
],
|
| 1155 |
+
"backend": "custom",
|
| 1156 |
+
"clean_up_tokenization_spaces": false,
|
| 1157 |
+
"eos_token": "</s>",
|
| 1158 |
+
"extra_ids": 0,
|
| 1159 |
+
"extra_special_tokens": [
|
| 1160 |
+
"<extra_id_0>",
|
| 1161 |
+
"<extra_id_1>",
|
| 1162 |
+
"<extra_id_2>",
|
| 1163 |
+
"<extra_id_3>",
|
| 1164 |
+
"<extra_id_4>",
|
| 1165 |
+
"<extra_id_5>",
|
| 1166 |
+
"<extra_id_6>",
|
| 1167 |
+
"<extra_id_7>",
|
| 1168 |
+
"<extra_id_8>",
|
| 1169 |
+
"<extra_id_9>",
|
| 1170 |
+
"<extra_id_10>",
|
| 1171 |
+
"<extra_id_11>",
|
| 1172 |
+
"<extra_id_12>",
|
| 1173 |
+
"<extra_id_13>",
|
| 1174 |
+
"<extra_id_14>",
|
| 1175 |
+
"<extra_id_15>",
|
| 1176 |
+
"<extra_id_16>",
|
| 1177 |
+
"<extra_id_17>",
|
| 1178 |
+
"<extra_id_18>",
|
| 1179 |
+
"<extra_id_19>",
|
| 1180 |
+
"<extra_id_20>",
|
| 1181 |
+
"<extra_id_21>",
|
| 1182 |
+
"<extra_id_22>",
|
| 1183 |
+
"<extra_id_23>",
|
| 1184 |
+
"<extra_id_24>",
|
| 1185 |
+
"<extra_id_25>",
|
| 1186 |
+
"<extra_id_26>",
|
| 1187 |
+
"<extra_id_27>",
|
| 1188 |
+
"<extra_id_28>",
|
| 1189 |
+
"<extra_id_29>",
|
| 1190 |
+
"<extra_id_30>",
|
| 1191 |
+
"<extra_id_31>",
|
| 1192 |
+
"<extra_id_32>",
|
| 1193 |
+
"<extra_id_33>",
|
| 1194 |
+
"<extra_id_34>",
|
| 1195 |
+
"<extra_id_35>",
|
| 1196 |
+
"<extra_id_36>",
|
| 1197 |
+
"<extra_id_37>",
|
| 1198 |
+
"<extra_id_38>",
|
| 1199 |
+
"<extra_id_39>",
|
| 1200 |
+
"<extra_id_40>",
|
| 1201 |
+
"<extra_id_41>",
|
| 1202 |
+
"<extra_id_42>",
|
| 1203 |
+
"<extra_id_43>",
|
| 1204 |
+
"<extra_id_44>",
|
| 1205 |
+
"<extra_id_45>",
|
| 1206 |
+
"<extra_id_46>",
|
| 1207 |
+
"<extra_id_47>",
|
| 1208 |
+
"<extra_id_48>",
|
| 1209 |
+
"<extra_id_49>",
|
| 1210 |
+
"<extra_id_50>",
|
| 1211 |
+
"<extra_id_51>",
|
| 1212 |
+
"<extra_id_52>",
|
| 1213 |
+
"<extra_id_53>",
|
| 1214 |
+
"<extra_id_54>",
|
| 1215 |
+
"<extra_id_55>",
|
| 1216 |
+
"<extra_id_56>",
|
| 1217 |
+
"<extra_id_57>",
|
| 1218 |
+
"<extra_id_58>",
|
| 1219 |
+
"<extra_id_59>",
|
| 1220 |
+
"<extra_id_60>",
|
| 1221 |
+
"<extra_id_61>",
|
| 1222 |
+
"<extra_id_62>",
|
| 1223 |
+
"<extra_id_63>",
|
| 1224 |
+
"<extra_id_64>",
|
| 1225 |
+
"<extra_id_65>",
|
| 1226 |
+
"<extra_id_66>",
|
| 1227 |
+
"<extra_id_67>",
|
| 1228 |
+
"<extra_id_68>",
|
| 1229 |
+
"<extra_id_69>",
|
| 1230 |
+
"<extra_id_70>",
|
| 1231 |
+
"<extra_id_71>",
|
| 1232 |
+
"<extra_id_72>",
|
| 1233 |
+
"<extra_id_73>",
|
| 1234 |
+
"<extra_id_74>",
|
| 1235 |
+
"<extra_id_75>",
|
| 1236 |
+
"<extra_id_76>",
|
| 1237 |
+
"<extra_id_77>",
|
| 1238 |
+
"<extra_id_78>",
|
| 1239 |
+
"<extra_id_79>",
|
| 1240 |
+
"<extra_id_80>",
|
| 1241 |
+
"<extra_id_81>",
|
| 1242 |
+
"<extra_id_82>",
|
| 1243 |
+
"<extra_id_83>",
|
| 1244 |
+
"<extra_id_84>",
|
| 1245 |
+
"<extra_id_85>",
|
| 1246 |
+
"<extra_id_86>",
|
| 1247 |
+
"<extra_id_87>",
|
| 1248 |
+
"<extra_id_88>",
|
| 1249 |
+
"<extra_id_89>",
|
| 1250 |
+
"<extra_id_90>",
|
| 1251 |
+
"<extra_id_91>",
|
| 1252 |
+
"<extra_id_92>",
|
| 1253 |
+
"<extra_id_93>",
|
| 1254 |
+
"<extra_id_94>",
|
| 1255 |
+
"<extra_id_95>",
|
| 1256 |
+
"<extra_id_96>",
|
| 1257 |
+
"<extra_id_97>",
|
| 1258 |
+
"<extra_id_98>",
|
| 1259 |
+
"<extra_id_99>",
|
| 1260 |
+
"<extra_id_100>",
|
| 1261 |
+
"<extra_id_101>",
|
| 1262 |
+
"<extra_id_102>",
|
| 1263 |
+
"<extra_id_103>",
|
| 1264 |
+
"<extra_id_104>",
|
| 1265 |
+
"<extra_id_105>",
|
| 1266 |
+
"<extra_id_106>",
|
| 1267 |
+
"<extra_id_107>",
|
| 1268 |
+
"<extra_id_108>",
|
| 1269 |
+
"<extra_id_109>",
|
| 1270 |
+
"<extra_id_110>",
|
| 1271 |
+
"<extra_id_111>",
|
| 1272 |
+
"<extra_id_112>",
|
| 1273 |
+
"<extra_id_113>",
|
| 1274 |
+
"<extra_id_114>",
|
| 1275 |
+
"<extra_id_115>",
|
| 1276 |
+
"<extra_id_116>",
|
| 1277 |
+
"<extra_id_117>",
|
| 1278 |
+
"<extra_id_118>",
|
| 1279 |
+
"<extra_id_119>",
|
| 1280 |
+
"<extra_id_120>",
|
| 1281 |
+
"<extra_id_121>",
|
| 1282 |
+
"<extra_id_122>",
|
| 1283 |
+
"<extra_id_123>",
|
| 1284 |
+
"<extra_id_124>"
|
| 1285 |
+
],
|
| 1286 |
+
"is_local": true,
|
| 1287 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 1288 |
+
"model_specific_special_tokens": {},
|
| 1289 |
+
"pad_token": "<pad>",
|
| 1290 |
+
"tokenizer_class": "ByT5Tokenizer",
|
| 1291 |
+
"unk_token": "<unk>"
|
| 1292 |
+
}
|
transformer/config.json
ADDED
|
@@ -0,0 +1,266 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_class_name": "GlmImageTransformer2DModel",
|
| 3 |
+
"_diffusers_version": "0.37.0.dev0",
|
| 4 |
+
"_name_or_path": "zai-org/GLM-Image",
|
| 5 |
+
"attention_head_dim": 128,
|
| 6 |
+
"condition_dim": 256,
|
| 7 |
+
"in_channels": 16,
|
| 8 |
+
"num_attention_heads": 32,
|
| 9 |
+
"num_layers": 30,
|
| 10 |
+
"out_channels": 16,
|
| 11 |
+
"patch_size": 2,
|
| 12 |
+
"prior_vq_quantizer_codebook_size": 16384,
|
| 13 |
+
"quantization_config": {
|
| 14 |
+
"add_skip_keys": true,
|
| 15 |
+
"dequantize_fp32": false,
|
| 16 |
+
"dynamic_loss_threshold": 0.01,
|
| 17 |
+
"group_size": 0,
|
| 18 |
+
"is_integer": true,
|
| 19 |
+
"is_training": false,
|
| 20 |
+
"modules_dtype_dict": {
|
| 21 |
+
"float5_e3m1fn": [
|
| 22 |
+
"transformer_blocks.23.norm1.linear.weight",
|
| 23 |
+
"transformer_blocks.24.norm1.linear.weight",
|
| 24 |
+
"transformer_blocks.25.norm1.linear.weight"
|
| 25 |
+
],
|
| 26 |
+
"int5": [
|
| 27 |
+
"transformer_blocks.0.attn1.to_k.weight",
|
| 28 |
+
"transformer_blocks.0.attn1.to_v.weight",
|
| 29 |
+
"transformer_blocks.0.ff.net.0.proj.weight",
|
| 30 |
+
"transformer_blocks.1.attn1.to_k.weight",
|
| 31 |
+
"transformer_blocks.1.attn1.to_q.weight",
|
| 32 |
+
"transformer_blocks.1.attn1.to_v.weight",
|
| 33 |
+
"transformer_blocks.1.ff.net.0.proj.weight",
|
| 34 |
+
"transformer_blocks.10.attn1.to_v.weight",
|
| 35 |
+
"transformer_blocks.2.attn1.to_k.weight",
|
| 36 |
+
"transformer_blocks.2.attn1.to_q.weight",
|
| 37 |
+
"transformer_blocks.2.attn1.to_v.weight",
|
| 38 |
+
"transformer_blocks.3.attn1.to_k.weight",
|
| 39 |
+
"transformer_blocks.3.attn1.to_q.weight",
|
| 40 |
+
"transformer_blocks.3.attn1.to_v.weight",
|
| 41 |
+
"transformer_blocks.3.ff.net.0.proj.weight",
|
| 42 |
+
"transformer_blocks.4.attn1.to_k.weight",
|
| 43 |
+
"transformer_blocks.4.attn1.to_q.weight",
|
| 44 |
+
"transformer_blocks.4.attn1.to_v.weight",
|
| 45 |
+
"transformer_blocks.5.attn1.to_v.weight",
|
| 46 |
+
"transformer_blocks.6.attn1.to_v.weight",
|
| 47 |
+
"transformer_blocks.7.attn1.to_v.weight",
|
| 48 |
+
"transformer_blocks.8.attn1.to_v.weight",
|
| 49 |
+
"transformer_blocks.9.attn1.to_v.weight",
|
| 50 |
+
"transformer_blocks.21.norm1.linear.weight",
|
| 51 |
+
"transformer_blocks.22.norm1.linear.weight",
|
| 52 |
+
"transformer_blocks.25.ff.net.2.weight",
|
| 53 |
+
"transformer_blocks.26.attn1.to_v.weight",
|
| 54 |
+
"transformer_blocks.26.ff.net.2.weight",
|
| 55 |
+
"transformer_blocks.26.norm1.linear.weight",
|
| 56 |
+
"transformer_blocks.27.attn1.to_v.weight",
|
| 57 |
+
"transformer_blocks.27.norm1.linear.weight",
|
| 58 |
+
"transformer_blocks.28.attn1.to_v.weight",
|
| 59 |
+
"transformer_blocks.29.attn1.to_v.weight"
|
| 60 |
+
],
|
| 61 |
+
"uint4": [
|
| 62 |
+
"transformer_blocks.0.attn1.to_out.0.weight",
|
| 63 |
+
"transformer_blocks.0.attn1.to_q.weight",
|
| 64 |
+
"transformer_blocks.0.ff.net.2.weight",
|
| 65 |
+
"transformer_blocks.1.attn1.to_out.0.weight",
|
| 66 |
+
"transformer_blocks.1.ff.net.2.weight",
|
| 67 |
+
"transformer_blocks.1.norm1.linear.weight",
|
| 68 |
+
"transformer_blocks.10.attn1.to_k.weight",
|
| 69 |
+
"transformer_blocks.10.attn1.to_out.0.weight",
|
| 70 |
+
"transformer_blocks.10.attn1.to_q.weight",
|
| 71 |
+
"transformer_blocks.10.norm1.linear.weight",
|
| 72 |
+
"transformer_blocks.2.attn1.to_out.0.weight",
|
| 73 |
+
"transformer_blocks.2.ff.net.0.proj.weight",
|
| 74 |
+
"transformer_blocks.2.ff.net.2.weight",
|
| 75 |
+
"transformer_blocks.2.norm1.linear.weight",
|
| 76 |
+
"transformer_blocks.3.attn1.to_out.0.weight",
|
| 77 |
+
"transformer_blocks.3.ff.net.2.weight",
|
| 78 |
+
"transformer_blocks.3.norm1.linear.weight",
|
| 79 |
+
"transformer_blocks.4.attn1.to_out.0.weight",
|
| 80 |
+
"transformer_blocks.4.ff.net.0.proj.weight",
|
| 81 |
+
"transformer_blocks.4.ff.net.2.weight",
|
| 82 |
+
"transformer_blocks.4.norm1.linear.weight",
|
| 83 |
+
"transformer_blocks.5.attn1.to_k.weight",
|
| 84 |
+
"transformer_blocks.5.attn1.to_out.0.weight",
|
| 85 |
+
"transformer_blocks.5.attn1.to_q.weight",
|
| 86 |
+
"transformer_blocks.5.ff.net.0.proj.weight",
|
| 87 |
+
"transformer_blocks.5.ff.net.2.weight",
|
| 88 |
+
"transformer_blocks.5.norm1.linear.weight",
|
| 89 |
+
"transformer_blocks.6.attn1.to_k.weight",
|
| 90 |
+
"transformer_blocks.6.attn1.to_out.0.weight",
|
| 91 |
+
"transformer_blocks.6.attn1.to_q.weight",
|
| 92 |
+
"transformer_blocks.6.ff.net.0.proj.weight",
|
| 93 |
+
"transformer_blocks.6.ff.net.2.weight",
|
| 94 |
+
"transformer_blocks.6.norm1.linear.weight",
|
| 95 |
+
"transformer_blocks.7.attn1.to_k.weight",
|
| 96 |
+
"transformer_blocks.7.attn1.to_out.0.weight",
|
| 97 |
+
"transformer_blocks.7.attn1.to_q.weight",
|
| 98 |
+
"transformer_blocks.7.ff.net.0.proj.weight",
|
| 99 |
+
"transformer_blocks.7.ff.net.2.weight",
|
| 100 |
+
"transformer_blocks.7.norm1.linear.weight",
|
| 101 |
+
"transformer_blocks.8.attn1.to_k.weight",
|
| 102 |
+
"transformer_blocks.8.attn1.to_out.0.weight",
|
| 103 |
+
"transformer_blocks.8.attn1.to_q.weight",
|
| 104 |
+
"transformer_blocks.8.ff.net.0.proj.weight",
|
| 105 |
+
"transformer_blocks.8.ff.net.2.weight",
|
| 106 |
+
"transformer_blocks.8.norm1.linear.weight",
|
| 107 |
+
"transformer_blocks.9.attn1.to_k.weight",
|
| 108 |
+
"transformer_blocks.9.attn1.to_out.0.weight",
|
| 109 |
+
"transformer_blocks.9.attn1.to_q.weight",
|
| 110 |
+
"transformer_blocks.9.ff.net.0.proj.weight",
|
| 111 |
+
"transformer_blocks.9.ff.net.2.weight",
|
| 112 |
+
"transformer_blocks.9.norm1.linear.weight",
|
| 113 |
+
"transformer_blocks.10.ff.net.0.proj.weight",
|
| 114 |
+
"transformer_blocks.10.ff.net.2.weight",
|
| 115 |
+
"transformer_blocks.11.attn1.to_k.weight",
|
| 116 |
+
"transformer_blocks.11.attn1.to_out.0.weight",
|
| 117 |
+
"transformer_blocks.11.attn1.to_q.weight",
|
| 118 |
+
"transformer_blocks.11.attn1.to_v.weight",
|
| 119 |
+
"transformer_blocks.11.ff.net.0.proj.weight",
|
| 120 |
+
"transformer_blocks.11.ff.net.2.weight",
|
| 121 |
+
"transformer_blocks.11.norm1.linear.weight",
|
| 122 |
+
"transformer_blocks.12.attn1.to_k.weight",
|
| 123 |
+
"transformer_blocks.12.attn1.to_out.0.weight",
|
| 124 |
+
"transformer_blocks.12.attn1.to_q.weight",
|
| 125 |
+
"transformer_blocks.12.attn1.to_v.weight",
|
| 126 |
+
"transformer_blocks.12.ff.net.0.proj.weight",
|
| 127 |
+
"transformer_blocks.12.ff.net.2.weight",
|
| 128 |
+
"transformer_blocks.12.norm1.linear.weight",
|
| 129 |
+
"transformer_blocks.13.attn1.to_k.weight",
|
| 130 |
+
"transformer_blocks.13.attn1.to_out.0.weight",
|
| 131 |
+
"transformer_blocks.13.attn1.to_q.weight",
|
| 132 |
+
"transformer_blocks.13.attn1.to_v.weight",
|
| 133 |
+
"transformer_blocks.13.ff.net.0.proj.weight",
|
| 134 |
+
"transformer_blocks.13.ff.net.2.weight",
|
| 135 |
+
"transformer_blocks.13.norm1.linear.weight",
|
| 136 |
+
"transformer_blocks.14.attn1.to_k.weight",
|
| 137 |
+
"transformer_blocks.14.attn1.to_out.0.weight",
|
| 138 |
+
"transformer_blocks.14.attn1.to_q.weight",
|
| 139 |
+
"transformer_blocks.14.attn1.to_v.weight",
|
| 140 |
+
"transformer_blocks.14.ff.net.0.proj.weight",
|
| 141 |
+
"transformer_blocks.14.ff.net.2.weight",
|
| 142 |
+
"transformer_blocks.14.norm1.linear.weight",
|
| 143 |
+
"transformer_blocks.15.attn1.to_k.weight",
|
| 144 |
+
"transformer_blocks.15.attn1.to_out.0.weight",
|
| 145 |
+
"transformer_blocks.15.attn1.to_q.weight",
|
| 146 |
+
"transformer_blocks.15.attn1.to_v.weight",
|
| 147 |
+
"transformer_blocks.15.ff.net.0.proj.weight",
|
| 148 |
+
"transformer_blocks.15.ff.net.2.weight",
|
| 149 |
+
"transformer_blocks.15.norm1.linear.weight",
|
| 150 |
+
"transformer_blocks.16.attn1.to_k.weight",
|
| 151 |
+
"transformer_blocks.16.attn1.to_out.0.weight",
|
| 152 |
+
"transformer_blocks.16.attn1.to_q.weight",
|
| 153 |
+
"transformer_blocks.16.attn1.to_v.weight",
|
| 154 |
+
"transformer_blocks.16.ff.net.0.proj.weight",
|
| 155 |
+
"transformer_blocks.16.ff.net.2.weight",
|
| 156 |
+
"transformer_blocks.16.norm1.linear.weight",
|
| 157 |
+
"transformer_blocks.17.attn1.to_k.weight",
|
| 158 |
+
"transformer_blocks.17.attn1.to_out.0.weight",
|
| 159 |
+
"transformer_blocks.17.attn1.to_q.weight",
|
| 160 |
+
"transformer_blocks.17.attn1.to_v.weight",
|
| 161 |
+
"transformer_blocks.17.ff.net.0.proj.weight",
|
| 162 |
+
"transformer_blocks.17.ff.net.2.weight",
|
| 163 |
+
"transformer_blocks.17.norm1.linear.weight",
|
| 164 |
+
"transformer_blocks.18.attn1.to_k.weight",
|
| 165 |
+
"transformer_blocks.18.attn1.to_out.0.weight",
|
| 166 |
+
"transformer_blocks.18.attn1.to_q.weight",
|
| 167 |
+
"transformer_blocks.18.attn1.to_v.weight",
|
| 168 |
+
"transformer_blocks.18.ff.net.0.proj.weight",
|
| 169 |
+
"transformer_blocks.18.ff.net.2.weight",
|
| 170 |
+
"transformer_blocks.18.norm1.linear.weight",
|
| 171 |
+
"transformer_blocks.19.attn1.to_k.weight",
|
| 172 |
+
"transformer_blocks.19.attn1.to_out.0.weight",
|
| 173 |
+
"transformer_blocks.19.attn1.to_q.weight",
|
| 174 |
+
"transformer_blocks.19.attn1.to_v.weight",
|
| 175 |
+
"transformer_blocks.19.ff.net.0.proj.weight",
|
| 176 |
+
"transformer_blocks.19.ff.net.2.weight",
|
| 177 |
+
"transformer_blocks.19.norm1.linear.weight",
|
| 178 |
+
"transformer_blocks.20.attn1.to_k.weight",
|
| 179 |
+
"transformer_blocks.20.attn1.to_out.0.weight",
|
| 180 |
+
"transformer_blocks.20.attn1.to_q.weight",
|
| 181 |
+
"transformer_blocks.20.attn1.to_v.weight",
|
| 182 |
+
"transformer_blocks.20.ff.net.0.proj.weight",
|
| 183 |
+
"transformer_blocks.20.ff.net.2.weight",
|
| 184 |
+
"transformer_blocks.20.norm1.linear.weight",
|
| 185 |
+
"transformer_blocks.21.attn1.to_k.weight",
|
| 186 |
+
"transformer_blocks.21.attn1.to_out.0.weight",
|
| 187 |
+
"transformer_blocks.21.attn1.to_q.weight",
|
| 188 |
+
"transformer_blocks.21.attn1.to_v.weight",
|
| 189 |
+
"transformer_blocks.21.ff.net.0.proj.weight",
|
| 190 |
+
"transformer_blocks.21.ff.net.2.weight",
|
| 191 |
+
"transformer_blocks.22.attn1.to_k.weight",
|
| 192 |
+
"transformer_blocks.22.attn1.to_out.0.weight",
|
| 193 |
+
"transformer_blocks.22.attn1.to_q.weight",
|
| 194 |
+
"transformer_blocks.22.attn1.to_v.weight",
|
| 195 |
+
"transformer_blocks.22.ff.net.0.proj.weight",
|
| 196 |
+
"transformer_blocks.22.ff.net.2.weight",
|
| 197 |
+
"transformer_blocks.23.attn1.to_k.weight",
|
| 198 |
+
"transformer_blocks.23.attn1.to_out.0.weight",
|
| 199 |
+
"transformer_blocks.23.attn1.to_q.weight",
|
| 200 |
+
"transformer_blocks.23.attn1.to_v.weight",
|
| 201 |
+
"transformer_blocks.23.ff.net.0.proj.weight",
|
| 202 |
+
"transformer_blocks.23.ff.net.2.weight",
|
| 203 |
+
"transformer_blocks.24.attn1.to_k.weight",
|
| 204 |
+
"transformer_blocks.24.attn1.to_out.0.weight",
|
| 205 |
+
"transformer_blocks.24.attn1.to_q.weight",
|
| 206 |
+
"transformer_blocks.24.attn1.to_v.weight",
|
| 207 |
+
"transformer_blocks.24.ff.net.0.proj.weight",
|
| 208 |
+
"transformer_blocks.24.ff.net.2.weight",
|
| 209 |
+
"transformer_blocks.25.attn1.to_k.weight",
|
| 210 |
+
"transformer_blocks.25.attn1.to_out.0.weight",
|
| 211 |
+
"transformer_blocks.25.attn1.to_q.weight",
|
| 212 |
+
"transformer_blocks.25.attn1.to_v.weight",
|
| 213 |
+
"transformer_blocks.25.ff.net.0.proj.weight",
|
| 214 |
+
"transformer_blocks.26.attn1.to_k.weight",
|
| 215 |
+
"transformer_blocks.26.attn1.to_out.0.weight",
|
| 216 |
+
"transformer_blocks.26.attn1.to_q.weight",
|
| 217 |
+
"transformer_blocks.26.ff.net.0.proj.weight",
|
| 218 |
+
"transformer_blocks.27.attn1.to_k.weight",
|
| 219 |
+
"transformer_blocks.27.attn1.to_out.0.weight",
|
| 220 |
+
"transformer_blocks.27.attn1.to_q.weight",
|
| 221 |
+
"transformer_blocks.27.ff.net.0.proj.weight",
|
| 222 |
+
"transformer_blocks.27.ff.net.2.weight",
|
| 223 |
+
"transformer_blocks.28.attn1.to_k.weight",
|
| 224 |
+
"transformer_blocks.28.attn1.to_out.0.weight",
|
| 225 |
+
"transformer_blocks.28.attn1.to_q.weight",
|
| 226 |
+
"transformer_blocks.28.ff.net.0.proj.weight",
|
| 227 |
+
"transformer_blocks.28.ff.net.2.weight",
|
| 228 |
+
"transformer_blocks.28.norm1.linear.weight",
|
| 229 |
+
"transformer_blocks.29.attn1.to_k.weight",
|
| 230 |
+
"transformer_blocks.29.attn1.to_out.0.weight",
|
| 231 |
+
"transformer_blocks.29.attn1.to_q.weight",
|
| 232 |
+
"transformer_blocks.29.ff.net.0.proj.weight",
|
| 233 |
+
"transformer_blocks.29.ff.net.2.weight",
|
| 234 |
+
"transformer_blocks.29.norm1.linear.weight"
|
| 235 |
+
]
|
| 236 |
+
},
|
| 237 |
+
"modules_to_not_convert": [
|
| 238 |
+
"transformer_blocks.0.norm1.linear.weight",
|
| 239 |
+
"time_condition_embed",
|
| 240 |
+
"image_projector",
|
| 241 |
+
"proj_out",
|
| 242 |
+
"prior_projector",
|
| 243 |
+
"glyph_projector",
|
| 244 |
+
"norm_out"
|
| 245 |
+
],
|
| 246 |
+
"non_blocking": false,
|
| 247 |
+
"quant_conv": false,
|
| 248 |
+
"quant_method": "sdnq",
|
| 249 |
+
"quantization_device": "cuda",
|
| 250 |
+
"quantized_matmul_dtype": null,
|
| 251 |
+
"return_device": "cuda",
|
| 252 |
+
"sdnq_version": "0.1.4",
|
| 253 |
+
"svd_rank": 32,
|
| 254 |
+
"svd_steps": 8,
|
| 255 |
+
"use_dynamic_quantization": true,
|
| 256 |
+
"use_grad_ckpt": true,
|
| 257 |
+
"use_quantized_matmul": false,
|
| 258 |
+
"use_quantized_matmul_conv": false,
|
| 259 |
+
"use_static_quantization": true,
|
| 260 |
+
"use_stochastic_rounding": false,
|
| 261 |
+
"use_svd": false,
|
| 262 |
+
"weights_dtype": "uint4"
|
| 263 |
+
},
|
| 264 |
+
"text_embed_dim": 1472,
|
| 265 |
+
"time_embed_dim": 512
|
| 266 |
+
}
|
transformer/diffusion_pytorch_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3a228933f18126ec5015f8d216cf5609a9cda9373ce662574abf159bb251843d
|
| 3 |
+
size 4193985424
|
transformer/quantization_config.json
ADDED
|
@@ -0,0 +1,251 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_skip_keys": false,
|
| 3 |
+
"dequantize_fp32": false,
|
| 4 |
+
"dynamic_loss_threshold": 0.01,
|
| 5 |
+
"group_size": 0,
|
| 6 |
+
"is_integer": true,
|
| 7 |
+
"is_training": false,
|
| 8 |
+
"modules_dtype_dict": {
|
| 9 |
+
"float5_e3m1fn": [
|
| 10 |
+
"transformer_blocks.23.norm1.linear.weight",
|
| 11 |
+
"transformer_blocks.24.norm1.linear.weight",
|
| 12 |
+
"transformer_blocks.25.norm1.linear.weight"
|
| 13 |
+
],
|
| 14 |
+
"int5": [
|
| 15 |
+
"transformer_blocks.0.attn1.to_k.weight",
|
| 16 |
+
"transformer_blocks.0.attn1.to_v.weight",
|
| 17 |
+
"transformer_blocks.0.ff.net.0.proj.weight",
|
| 18 |
+
"transformer_blocks.1.attn1.to_k.weight",
|
| 19 |
+
"transformer_blocks.1.attn1.to_q.weight",
|
| 20 |
+
"transformer_blocks.1.attn1.to_v.weight",
|
| 21 |
+
"transformer_blocks.1.ff.net.0.proj.weight",
|
| 22 |
+
"transformer_blocks.10.attn1.to_v.weight",
|
| 23 |
+
"transformer_blocks.2.attn1.to_k.weight",
|
| 24 |
+
"transformer_blocks.2.attn1.to_q.weight",
|
| 25 |
+
"transformer_blocks.2.attn1.to_v.weight",
|
| 26 |
+
"transformer_blocks.3.attn1.to_k.weight",
|
| 27 |
+
"transformer_blocks.3.attn1.to_q.weight",
|
| 28 |
+
"transformer_blocks.3.attn1.to_v.weight",
|
| 29 |
+
"transformer_blocks.3.ff.net.0.proj.weight",
|
| 30 |
+
"transformer_blocks.4.attn1.to_k.weight",
|
| 31 |
+
"transformer_blocks.4.attn1.to_q.weight",
|
| 32 |
+
"transformer_blocks.4.attn1.to_v.weight",
|
| 33 |
+
"transformer_blocks.5.attn1.to_v.weight",
|
| 34 |
+
"transformer_blocks.6.attn1.to_v.weight",
|
| 35 |
+
"transformer_blocks.7.attn1.to_v.weight",
|
| 36 |
+
"transformer_blocks.8.attn1.to_v.weight",
|
| 37 |
+
"transformer_blocks.9.attn1.to_v.weight",
|
| 38 |
+
"transformer_blocks.21.norm1.linear.weight",
|
| 39 |
+
"transformer_blocks.22.norm1.linear.weight",
|
| 40 |
+
"transformer_blocks.25.ff.net.2.weight",
|
| 41 |
+
"transformer_blocks.26.attn1.to_v.weight",
|
| 42 |
+
"transformer_blocks.26.ff.net.2.weight",
|
| 43 |
+
"transformer_blocks.26.norm1.linear.weight",
|
| 44 |
+
"transformer_blocks.27.attn1.to_v.weight",
|
| 45 |
+
"transformer_blocks.27.norm1.linear.weight",
|
| 46 |
+
"transformer_blocks.28.attn1.to_v.weight",
|
| 47 |
+
"transformer_blocks.29.attn1.to_v.weight"
|
| 48 |
+
],
|
| 49 |
+
"uint4": [
|
| 50 |
+
"transformer_blocks.0.attn1.to_out.0.weight",
|
| 51 |
+
"transformer_blocks.0.attn1.to_q.weight",
|
| 52 |
+
"transformer_blocks.0.ff.net.2.weight",
|
| 53 |
+
"transformer_blocks.1.attn1.to_out.0.weight",
|
| 54 |
+
"transformer_blocks.1.ff.net.2.weight",
|
| 55 |
+
"transformer_blocks.1.norm1.linear.weight",
|
| 56 |
+
"transformer_blocks.10.attn1.to_k.weight",
|
| 57 |
+
"transformer_blocks.10.attn1.to_out.0.weight",
|
| 58 |
+
"transformer_blocks.10.attn1.to_q.weight",
|
| 59 |
+
"transformer_blocks.10.norm1.linear.weight",
|
| 60 |
+
"transformer_blocks.2.attn1.to_out.0.weight",
|
| 61 |
+
"transformer_blocks.2.ff.net.0.proj.weight",
|
| 62 |
+
"transformer_blocks.2.ff.net.2.weight",
|
| 63 |
+
"transformer_blocks.2.norm1.linear.weight",
|
| 64 |
+
"transformer_blocks.3.attn1.to_out.0.weight",
|
| 65 |
+
"transformer_blocks.3.ff.net.2.weight",
|
| 66 |
+
"transformer_blocks.3.norm1.linear.weight",
|
| 67 |
+
"transformer_blocks.4.attn1.to_out.0.weight",
|
| 68 |
+
"transformer_blocks.4.ff.net.0.proj.weight",
|
| 69 |
+
"transformer_blocks.4.ff.net.2.weight",
|
| 70 |
+
"transformer_blocks.4.norm1.linear.weight",
|
| 71 |
+
"transformer_blocks.5.attn1.to_k.weight",
|
| 72 |
+
"transformer_blocks.5.attn1.to_out.0.weight",
|
| 73 |
+
"transformer_blocks.5.attn1.to_q.weight",
|
| 74 |
+
"transformer_blocks.5.ff.net.0.proj.weight",
|
| 75 |
+
"transformer_blocks.5.ff.net.2.weight",
|
| 76 |
+
"transformer_blocks.5.norm1.linear.weight",
|
| 77 |
+
"transformer_blocks.6.attn1.to_k.weight",
|
| 78 |
+
"transformer_blocks.6.attn1.to_out.0.weight",
|
| 79 |
+
"transformer_blocks.6.attn1.to_q.weight",
|
| 80 |
+
"transformer_blocks.6.ff.net.0.proj.weight",
|
| 81 |
+
"transformer_blocks.6.ff.net.2.weight",
|
| 82 |
+
"transformer_blocks.6.norm1.linear.weight",
|
| 83 |
+
"transformer_blocks.7.attn1.to_k.weight",
|
| 84 |
+
"transformer_blocks.7.attn1.to_out.0.weight",
|
| 85 |
+
"transformer_blocks.7.attn1.to_q.weight",
|
| 86 |
+
"transformer_blocks.7.ff.net.0.proj.weight",
|
| 87 |
+
"transformer_blocks.7.ff.net.2.weight",
|
| 88 |
+
"transformer_blocks.7.norm1.linear.weight",
|
| 89 |
+
"transformer_blocks.8.attn1.to_k.weight",
|
| 90 |
+
"transformer_blocks.8.attn1.to_out.0.weight",
|
| 91 |
+
"transformer_blocks.8.attn1.to_q.weight",
|
| 92 |
+
"transformer_blocks.8.ff.net.0.proj.weight",
|
| 93 |
+
"transformer_blocks.8.ff.net.2.weight",
|
| 94 |
+
"transformer_blocks.8.norm1.linear.weight",
|
| 95 |
+
"transformer_blocks.9.attn1.to_k.weight",
|
| 96 |
+
"transformer_blocks.9.attn1.to_out.0.weight",
|
| 97 |
+
"transformer_blocks.9.attn1.to_q.weight",
|
| 98 |
+
"transformer_blocks.9.ff.net.0.proj.weight",
|
| 99 |
+
"transformer_blocks.9.ff.net.2.weight",
|
| 100 |
+
"transformer_blocks.9.norm1.linear.weight",
|
| 101 |
+
"transformer_blocks.10.ff.net.0.proj.weight",
|
| 102 |
+
"transformer_blocks.10.ff.net.2.weight",
|
| 103 |
+
"transformer_blocks.11.attn1.to_k.weight",
|
| 104 |
+
"transformer_blocks.11.attn1.to_out.0.weight",
|
| 105 |
+
"transformer_blocks.11.attn1.to_q.weight",
|
| 106 |
+
"transformer_blocks.11.attn1.to_v.weight",
|
| 107 |
+
"transformer_blocks.11.ff.net.0.proj.weight",
|
| 108 |
+
"transformer_blocks.11.ff.net.2.weight",
|
| 109 |
+
"transformer_blocks.11.norm1.linear.weight",
|
| 110 |
+
"transformer_blocks.12.attn1.to_k.weight",
|
| 111 |
+
"transformer_blocks.12.attn1.to_out.0.weight",
|
| 112 |
+
"transformer_blocks.12.attn1.to_q.weight",
|
| 113 |
+
"transformer_blocks.12.attn1.to_v.weight",
|
| 114 |
+
"transformer_blocks.12.ff.net.0.proj.weight",
|
| 115 |
+
"transformer_blocks.12.ff.net.2.weight",
|
| 116 |
+
"transformer_blocks.12.norm1.linear.weight",
|
| 117 |
+
"transformer_blocks.13.attn1.to_k.weight",
|
| 118 |
+
"transformer_blocks.13.attn1.to_out.0.weight",
|
| 119 |
+
"transformer_blocks.13.attn1.to_q.weight",
|
| 120 |
+
"transformer_blocks.13.attn1.to_v.weight",
|
| 121 |
+
"transformer_blocks.13.ff.net.0.proj.weight",
|
| 122 |
+
"transformer_blocks.13.ff.net.2.weight",
|
| 123 |
+
"transformer_blocks.13.norm1.linear.weight",
|
| 124 |
+
"transformer_blocks.14.attn1.to_k.weight",
|
| 125 |
+
"transformer_blocks.14.attn1.to_out.0.weight",
|
| 126 |
+
"transformer_blocks.14.attn1.to_q.weight",
|
| 127 |
+
"transformer_blocks.14.attn1.to_v.weight",
|
| 128 |
+
"transformer_blocks.14.ff.net.0.proj.weight",
|
| 129 |
+
"transformer_blocks.14.ff.net.2.weight",
|
| 130 |
+
"transformer_blocks.14.norm1.linear.weight",
|
| 131 |
+
"transformer_blocks.15.attn1.to_k.weight",
|
| 132 |
+
"transformer_blocks.15.attn1.to_out.0.weight",
|
| 133 |
+
"transformer_blocks.15.attn1.to_q.weight",
|
| 134 |
+
"transformer_blocks.15.attn1.to_v.weight",
|
| 135 |
+
"transformer_blocks.15.ff.net.0.proj.weight",
|
| 136 |
+
"transformer_blocks.15.ff.net.2.weight",
|
| 137 |
+
"transformer_blocks.15.norm1.linear.weight",
|
| 138 |
+
"transformer_blocks.16.attn1.to_k.weight",
|
| 139 |
+
"transformer_blocks.16.attn1.to_out.0.weight",
|
| 140 |
+
"transformer_blocks.16.attn1.to_q.weight",
|
| 141 |
+
"transformer_blocks.16.attn1.to_v.weight",
|
| 142 |
+
"transformer_blocks.16.ff.net.0.proj.weight",
|
| 143 |
+
"transformer_blocks.16.ff.net.2.weight",
|
| 144 |
+
"transformer_blocks.16.norm1.linear.weight",
|
| 145 |
+
"transformer_blocks.17.attn1.to_k.weight",
|
| 146 |
+
"transformer_blocks.17.attn1.to_out.0.weight",
|
| 147 |
+
"transformer_blocks.17.attn1.to_q.weight",
|
| 148 |
+
"transformer_blocks.17.attn1.to_v.weight",
|
| 149 |
+
"transformer_blocks.17.ff.net.0.proj.weight",
|
| 150 |
+
"transformer_blocks.17.ff.net.2.weight",
|
| 151 |
+
"transformer_blocks.17.norm1.linear.weight",
|
| 152 |
+
"transformer_blocks.18.attn1.to_k.weight",
|
| 153 |
+
"transformer_blocks.18.attn1.to_out.0.weight",
|
| 154 |
+
"transformer_blocks.18.attn1.to_q.weight",
|
| 155 |
+
"transformer_blocks.18.attn1.to_v.weight",
|
| 156 |
+
"transformer_blocks.18.ff.net.0.proj.weight",
|
| 157 |
+
"transformer_blocks.18.ff.net.2.weight",
|
| 158 |
+
"transformer_blocks.18.norm1.linear.weight",
|
| 159 |
+
"transformer_blocks.19.attn1.to_k.weight",
|
| 160 |
+
"transformer_blocks.19.attn1.to_out.0.weight",
|
| 161 |
+
"transformer_blocks.19.attn1.to_q.weight",
|
| 162 |
+
"transformer_blocks.19.attn1.to_v.weight",
|
| 163 |
+
"transformer_blocks.19.ff.net.0.proj.weight",
|
| 164 |
+
"transformer_blocks.19.ff.net.2.weight",
|
| 165 |
+
"transformer_blocks.19.norm1.linear.weight",
|
| 166 |
+
"transformer_blocks.20.attn1.to_k.weight",
|
| 167 |
+
"transformer_blocks.20.attn1.to_out.0.weight",
|
| 168 |
+
"transformer_blocks.20.attn1.to_q.weight",
|
| 169 |
+
"transformer_blocks.20.attn1.to_v.weight",
|
| 170 |
+
"transformer_blocks.20.ff.net.0.proj.weight",
|
| 171 |
+
"transformer_blocks.20.ff.net.2.weight",
|
| 172 |
+
"transformer_blocks.20.norm1.linear.weight",
|
| 173 |
+
"transformer_blocks.21.attn1.to_k.weight",
|
| 174 |
+
"transformer_blocks.21.attn1.to_out.0.weight",
|
| 175 |
+
"transformer_blocks.21.attn1.to_q.weight",
|
| 176 |
+
"transformer_blocks.21.attn1.to_v.weight",
|
| 177 |
+
"transformer_blocks.21.ff.net.0.proj.weight",
|
| 178 |
+
"transformer_blocks.21.ff.net.2.weight",
|
| 179 |
+
"transformer_blocks.22.attn1.to_k.weight",
|
| 180 |
+
"transformer_blocks.22.attn1.to_out.0.weight",
|
| 181 |
+
"transformer_blocks.22.attn1.to_q.weight",
|
| 182 |
+
"transformer_blocks.22.attn1.to_v.weight",
|
| 183 |
+
"transformer_blocks.22.ff.net.0.proj.weight",
|
| 184 |
+
"transformer_blocks.22.ff.net.2.weight",
|
| 185 |
+
"transformer_blocks.23.attn1.to_k.weight",
|
| 186 |
+
"transformer_blocks.23.attn1.to_out.0.weight",
|
| 187 |
+
"transformer_blocks.23.attn1.to_q.weight",
|
| 188 |
+
"transformer_blocks.23.attn1.to_v.weight",
|
| 189 |
+
"transformer_blocks.23.ff.net.0.proj.weight",
|
| 190 |
+
"transformer_blocks.23.ff.net.2.weight",
|
| 191 |
+
"transformer_blocks.24.attn1.to_k.weight",
|
| 192 |
+
"transformer_blocks.24.attn1.to_out.0.weight",
|
| 193 |
+
"transformer_blocks.24.attn1.to_q.weight",
|
| 194 |
+
"transformer_blocks.24.attn1.to_v.weight",
|
| 195 |
+
"transformer_blocks.24.ff.net.0.proj.weight",
|
| 196 |
+
"transformer_blocks.24.ff.net.2.weight",
|
| 197 |
+
"transformer_blocks.25.attn1.to_k.weight",
|
| 198 |
+
"transformer_blocks.25.attn1.to_out.0.weight",
|
| 199 |
+
"transformer_blocks.25.attn1.to_q.weight",
|
| 200 |
+
"transformer_blocks.25.attn1.to_v.weight",
|
| 201 |
+
"transformer_blocks.25.ff.net.0.proj.weight",
|
| 202 |
+
"transformer_blocks.26.attn1.to_k.weight",
|
| 203 |
+
"transformer_blocks.26.attn1.to_out.0.weight",
|
| 204 |
+
"transformer_blocks.26.attn1.to_q.weight",
|
| 205 |
+
"transformer_blocks.26.ff.net.0.proj.weight",
|
| 206 |
+
"transformer_blocks.27.attn1.to_k.weight",
|
| 207 |
+
"transformer_blocks.27.attn1.to_out.0.weight",
|
| 208 |
+
"transformer_blocks.27.attn1.to_q.weight",
|
| 209 |
+
"transformer_blocks.27.ff.net.0.proj.weight",
|
| 210 |
+
"transformer_blocks.27.ff.net.2.weight",
|
| 211 |
+
"transformer_blocks.28.attn1.to_k.weight",
|
| 212 |
+
"transformer_blocks.28.attn1.to_out.0.weight",
|
| 213 |
+
"transformer_blocks.28.attn1.to_q.weight",
|
| 214 |
+
"transformer_blocks.28.ff.net.0.proj.weight",
|
| 215 |
+
"transformer_blocks.28.ff.net.2.weight",
|
| 216 |
+
"transformer_blocks.28.norm1.linear.weight",
|
| 217 |
+
"transformer_blocks.29.attn1.to_k.weight",
|
| 218 |
+
"transformer_blocks.29.attn1.to_out.0.weight",
|
| 219 |
+
"transformer_blocks.29.attn1.to_q.weight",
|
| 220 |
+
"transformer_blocks.29.ff.net.0.proj.weight",
|
| 221 |
+
"transformer_blocks.29.ff.net.2.weight",
|
| 222 |
+
"transformer_blocks.29.norm1.linear.weight"
|
| 223 |
+
]
|
| 224 |
+
},
|
| 225 |
+
"modules_to_not_convert": [
|
| 226 |
+
"transformer_blocks.0.norm1.linear.weight",
|
| 227 |
+
"time_condition_embed",
|
| 228 |
+
"image_projector",
|
| 229 |
+
"proj_out",
|
| 230 |
+
"prior_projector",
|
| 231 |
+
"glyph_projector",
|
| 232 |
+
"norm_out"
|
| 233 |
+
],
|
| 234 |
+
"non_blocking": false,
|
| 235 |
+
"quant_conv": false,
|
| 236 |
+
"quant_method": "sdnq",
|
| 237 |
+
"quantization_device": null,
|
| 238 |
+
"quantized_matmul_dtype": null,
|
| 239 |
+
"return_device": null,
|
| 240 |
+
"sdnq_version": "0.1.4",
|
| 241 |
+
"svd_rank": 32,
|
| 242 |
+
"svd_steps": 8,
|
| 243 |
+
"use_dynamic_quantization": true,
|
| 244 |
+
"use_grad_ckpt": true,
|
| 245 |
+
"use_quantized_matmul": false,
|
| 246 |
+
"use_quantized_matmul_conv": false,
|
| 247 |
+
"use_static_quantization": true,
|
| 248 |
+
"use_stochastic_rounding": false,
|
| 249 |
+
"use_svd": false,
|
| 250 |
+
"weights_dtype": "uint4"
|
| 251 |
+
}
|
vae/config.json
ADDED
|
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_class_name": "AutoencoderKL",
|
| 3 |
+
"_diffusers_version": "0.37.0.dev0",
|
| 4 |
+
"_name_or_path": "/home/disty/.cache/huggingface/hub/models--zai-org--GLM-Image/snapshots/a32e429538178757f852fd54b52a5e1681da1e35/vae",
|
| 5 |
+
"act_fn": "silu",
|
| 6 |
+
"block_out_channels": [
|
| 7 |
+
128,
|
| 8 |
+
512,
|
| 9 |
+
1024,
|
| 10 |
+
1024
|
| 11 |
+
],
|
| 12 |
+
"down_block_types": [
|
| 13 |
+
"DownEncoderBlock2D",
|
| 14 |
+
"DownEncoderBlock2D",
|
| 15 |
+
"DownEncoderBlock2D",
|
| 16 |
+
"DownEncoderBlock2D"
|
| 17 |
+
],
|
| 18 |
+
"force_upcast": true,
|
| 19 |
+
"in_channels": 3,
|
| 20 |
+
"latent_channels": 16,
|
| 21 |
+
"latents_mean": [
|
| 22 |
+
-0.2080078125,
|
| 23 |
+
1.875,
|
| 24 |
+
-0.470703125,
|
| 25 |
+
-1.265625,
|
| 26 |
+
-1.421875,
|
| 27 |
+
0.77734375,
|
| 28 |
+
-0.3671875,
|
| 29 |
+
-0.9453125,
|
| 30 |
+
0.318359375,
|
| 31 |
+
0.7734375,
|
| 32 |
+
-0.1884765625,
|
| 33 |
+
-0.022216796875,
|
| 34 |
+
-0.220703125,
|
| 35 |
+
-1.59375,
|
| 36 |
+
-0.81640625,
|
| 37 |
+
-0.255859375
|
| 38 |
+
],
|
| 39 |
+
"latents_std": [
|
| 40 |
+
3.0625,
|
| 41 |
+
2.203125,
|
| 42 |
+
2.265625,
|
| 43 |
+
4.84375,
|
| 44 |
+
2.5,
|
| 45 |
+
3.9375,
|
| 46 |
+
2.203125,
|
| 47 |
+
3.03125,
|
| 48 |
+
2.1875,
|
| 49 |
+
2.046875,
|
| 50 |
+
2.71875,
|
| 51 |
+
2.390625,
|
| 52 |
+
2.390625,
|
| 53 |
+
2.453125,
|
| 54 |
+
2.25,
|
| 55 |
+
2.15625
|
| 56 |
+
],
|
| 57 |
+
"layers_per_block": 3,
|
| 58 |
+
"mid_block_add_attention": false,
|
| 59 |
+
"norm_num_groups": 32,
|
| 60 |
+
"out_channels": 3,
|
| 61 |
+
"sample_size": 1024,
|
| 62 |
+
"scaling_factor": 0.18215,
|
| 63 |
+
"shift_factor": null,
|
| 64 |
+
"up_block_types": [
|
| 65 |
+
"UpDecoderBlock2D",
|
| 66 |
+
"UpDecoderBlock2D",
|
| 67 |
+
"UpDecoderBlock2D",
|
| 68 |
+
"UpDecoderBlock2D"
|
| 69 |
+
],
|
| 70 |
+
"use_post_quant_conv": false,
|
| 71 |
+
"use_quant_conv": false
|
| 72 |
+
}
|
vae/diffusion_pytorch_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a80a129b649cdb6d74d9c2b5bb060f3b882b7826b63a732306a7d3393deedbc0
|
| 3 |
+
size 812234342
|
vision_language_encoder/config.json
ADDED
|
@@ -0,0 +1,505 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"GlmImageForConditionalGeneration"
|
| 4 |
+
],
|
| 5 |
+
"dtype": "bfloat16",
|
| 6 |
+
"image_end_token_id": 16385,
|
| 7 |
+
"image_start_token_id": 16384,
|
| 8 |
+
"image_token_id": 167855,
|
| 9 |
+
"model_type": "glm_image",
|
| 10 |
+
"quantization_config": {
|
| 11 |
+
"add_skip_keys": false,
|
| 12 |
+
"dequantize_fp32": false,
|
| 13 |
+
"dynamic_loss_threshold": 0.01,
|
| 14 |
+
"group_size": 0,
|
| 15 |
+
"is_integer": true,
|
| 16 |
+
"is_training": false,
|
| 17 |
+
"modules_dtype_dict": {
|
| 18 |
+
"int5": [
|
| 19 |
+
"model.language_model.layers.28.self_attn.v_proj.weight",
|
| 20 |
+
"model.language_model.layers.29.self_attn.v_proj.weight",
|
| 21 |
+
"model.language_model.layers.32.self_attn.v_proj.weight",
|
| 22 |
+
"model.language_model.layers.33.self_attn.v_proj.weight",
|
| 23 |
+
"model.visual.blocks.0.attn.qkv.weight",
|
| 24 |
+
"model.visual.blocks.0.mlp.fc2.weight",
|
| 25 |
+
"model.visual.blocks.1.attn.qkv.weight",
|
| 26 |
+
"model.visual.blocks.1.mlp.fc1.weight",
|
| 27 |
+
"model.visual.blocks.1.mlp.fc2.weight",
|
| 28 |
+
"model.visual.blocks.2.attn.qkv.weight",
|
| 29 |
+
"model.visual.blocks.2.mlp.fc2.weight",
|
| 30 |
+
"model.visual.blocks.3.attn.qkv.weight",
|
| 31 |
+
"model.visual.blocks.3.mlp.fc2.weight",
|
| 32 |
+
"model.visual.blocks.4.mlp.fc2.weight",
|
| 33 |
+
"model.visual.blocks.5.mlp.fc2.weight",
|
| 34 |
+
"model.visual.blocks.6.mlp.fc2.weight",
|
| 35 |
+
"model.visual.blocks.7.mlp.fc2.weight",
|
| 36 |
+
"model.visual.blocks.8.mlp.fc2.weight",
|
| 37 |
+
"model.visual.blocks.9.mlp.fc2.weight"
|
| 38 |
+
],
|
| 39 |
+
"uint4": [
|
| 40 |
+
"model.language_model.layers.0.mlp.down_proj.weight",
|
| 41 |
+
"model.language_model.layers.0.mlp.gate_up_proj.weight",
|
| 42 |
+
"model.language_model.layers.0.self_attn.k_proj.weight",
|
| 43 |
+
"model.language_model.layers.0.self_attn.o_proj.weight",
|
| 44 |
+
"model.language_model.layers.0.self_attn.q_proj.weight",
|
| 45 |
+
"model.language_model.layers.0.self_attn.v_proj.weight",
|
| 46 |
+
"model.language_model.layers.1.mlp.down_proj.weight",
|
| 47 |
+
"model.language_model.layers.1.mlp.gate_up_proj.weight",
|
| 48 |
+
"model.language_model.layers.1.self_attn.k_proj.weight",
|
| 49 |
+
"model.language_model.layers.1.self_attn.o_proj.weight",
|
| 50 |
+
"model.language_model.layers.1.self_attn.q_proj.weight",
|
| 51 |
+
"model.language_model.layers.1.self_attn.v_proj.weight",
|
| 52 |
+
"model.language_model.layers.2.mlp.down_proj.weight",
|
| 53 |
+
"model.language_model.layers.2.mlp.gate_up_proj.weight",
|
| 54 |
+
"model.language_model.layers.2.self_attn.k_proj.weight",
|
| 55 |
+
"model.language_model.layers.2.self_attn.o_proj.weight",
|
| 56 |
+
"model.language_model.layers.2.self_attn.q_proj.weight",
|
| 57 |
+
"model.language_model.layers.2.self_attn.v_proj.weight",
|
| 58 |
+
"model.language_model.layers.3.mlp.down_proj.weight",
|
| 59 |
+
"model.language_model.layers.3.mlp.gate_up_proj.weight",
|
| 60 |
+
"model.language_model.layers.3.self_attn.k_proj.weight",
|
| 61 |
+
"model.language_model.layers.3.self_attn.o_proj.weight",
|
| 62 |
+
"model.language_model.layers.3.self_attn.q_proj.weight",
|
| 63 |
+
"model.language_model.layers.3.self_attn.v_proj.weight",
|
| 64 |
+
"model.language_model.layers.4.mlp.down_proj.weight",
|
| 65 |
+
"model.language_model.layers.4.mlp.gate_up_proj.weight",
|
| 66 |
+
"model.language_model.layers.4.self_attn.k_proj.weight",
|
| 67 |
+
"model.language_model.layers.4.self_attn.o_proj.weight",
|
| 68 |
+
"model.language_model.layers.4.self_attn.q_proj.weight",
|
| 69 |
+
"model.language_model.layers.4.self_attn.v_proj.weight",
|
| 70 |
+
"model.language_model.layers.5.mlp.down_proj.weight",
|
| 71 |
+
"model.language_model.layers.5.mlp.gate_up_proj.weight",
|
| 72 |
+
"model.language_model.layers.5.self_attn.k_proj.weight",
|
| 73 |
+
"model.language_model.layers.5.self_attn.o_proj.weight",
|
| 74 |
+
"model.language_model.layers.5.self_attn.q_proj.weight",
|
| 75 |
+
"model.language_model.layers.5.self_attn.v_proj.weight",
|
| 76 |
+
"model.language_model.layers.6.mlp.down_proj.weight",
|
| 77 |
+
"model.language_model.layers.6.mlp.gate_up_proj.weight",
|
| 78 |
+
"model.language_model.layers.6.self_attn.k_proj.weight",
|
| 79 |
+
"model.language_model.layers.6.self_attn.o_proj.weight",
|
| 80 |
+
"model.language_model.layers.6.self_attn.q_proj.weight",
|
| 81 |
+
"model.language_model.layers.6.self_attn.v_proj.weight",
|
| 82 |
+
"model.language_model.layers.7.mlp.down_proj.weight",
|
| 83 |
+
"model.language_model.layers.7.mlp.gate_up_proj.weight",
|
| 84 |
+
"model.language_model.layers.7.self_attn.k_proj.weight",
|
| 85 |
+
"model.language_model.layers.7.self_attn.o_proj.weight",
|
| 86 |
+
"model.language_model.layers.7.self_attn.q_proj.weight",
|
| 87 |
+
"model.language_model.layers.7.self_attn.v_proj.weight",
|
| 88 |
+
"model.language_model.layers.8.mlp.down_proj.weight",
|
| 89 |
+
"model.language_model.layers.8.mlp.gate_up_proj.weight",
|
| 90 |
+
"model.language_model.layers.8.self_attn.k_proj.weight",
|
| 91 |
+
"model.language_model.layers.8.self_attn.o_proj.weight",
|
| 92 |
+
"model.language_model.layers.8.self_attn.q_proj.weight",
|
| 93 |
+
"model.language_model.layers.8.self_attn.v_proj.weight",
|
| 94 |
+
"model.language_model.layers.9.mlp.down_proj.weight",
|
| 95 |
+
"model.language_model.layers.9.mlp.gate_up_proj.weight",
|
| 96 |
+
"model.language_model.layers.9.self_attn.k_proj.weight",
|
| 97 |
+
"model.language_model.layers.9.self_attn.o_proj.weight",
|
| 98 |
+
"model.language_model.layers.9.self_attn.q_proj.weight",
|
| 99 |
+
"model.language_model.layers.9.self_attn.v_proj.weight",
|
| 100 |
+
"model.language_model.layers.10.mlp.down_proj.weight",
|
| 101 |
+
"model.language_model.layers.10.mlp.gate_up_proj.weight",
|
| 102 |
+
"model.language_model.layers.10.self_attn.k_proj.weight",
|
| 103 |
+
"model.language_model.layers.10.self_attn.o_proj.weight",
|
| 104 |
+
"model.language_model.layers.10.self_attn.q_proj.weight",
|
| 105 |
+
"model.language_model.layers.10.self_attn.v_proj.weight",
|
| 106 |
+
"model.language_model.layers.11.mlp.down_proj.weight",
|
| 107 |
+
"model.language_model.layers.11.mlp.gate_up_proj.weight",
|
| 108 |
+
"model.language_model.layers.11.self_attn.k_proj.weight",
|
| 109 |
+
"model.language_model.layers.11.self_attn.o_proj.weight",
|
| 110 |
+
"model.language_model.layers.11.self_attn.q_proj.weight",
|
| 111 |
+
"model.language_model.layers.11.self_attn.v_proj.weight",
|
| 112 |
+
"model.language_model.layers.12.mlp.down_proj.weight",
|
| 113 |
+
"model.language_model.layers.12.mlp.gate_up_proj.weight",
|
| 114 |
+
"model.language_model.layers.12.self_attn.k_proj.weight",
|
| 115 |
+
"model.language_model.layers.12.self_attn.o_proj.weight",
|
| 116 |
+
"model.language_model.layers.12.self_attn.q_proj.weight",
|
| 117 |
+
"model.language_model.layers.12.self_attn.v_proj.weight",
|
| 118 |
+
"model.language_model.layers.13.mlp.down_proj.weight",
|
| 119 |
+
"model.language_model.layers.13.mlp.gate_up_proj.weight",
|
| 120 |
+
"model.language_model.layers.13.self_attn.k_proj.weight",
|
| 121 |
+
"model.language_model.layers.13.self_attn.o_proj.weight",
|
| 122 |
+
"model.language_model.layers.13.self_attn.q_proj.weight",
|
| 123 |
+
"model.language_model.layers.13.self_attn.v_proj.weight",
|
| 124 |
+
"model.language_model.layers.14.mlp.down_proj.weight",
|
| 125 |
+
"model.language_model.layers.14.mlp.gate_up_proj.weight",
|
| 126 |
+
"model.language_model.layers.14.self_attn.k_proj.weight",
|
| 127 |
+
"model.language_model.layers.14.self_attn.o_proj.weight",
|
| 128 |
+
"model.language_model.layers.14.self_attn.q_proj.weight",
|
| 129 |
+
"model.language_model.layers.14.self_attn.v_proj.weight",
|
| 130 |
+
"model.language_model.layers.15.mlp.down_proj.weight",
|
| 131 |
+
"model.language_model.layers.15.mlp.gate_up_proj.weight",
|
| 132 |
+
"model.language_model.layers.15.self_attn.k_proj.weight",
|
| 133 |
+
"model.language_model.layers.15.self_attn.o_proj.weight",
|
| 134 |
+
"model.language_model.layers.15.self_attn.q_proj.weight",
|
| 135 |
+
"model.language_model.layers.15.self_attn.v_proj.weight",
|
| 136 |
+
"model.language_model.layers.16.mlp.down_proj.weight",
|
| 137 |
+
"model.language_model.layers.16.mlp.gate_up_proj.weight",
|
| 138 |
+
"model.language_model.layers.16.self_attn.k_proj.weight",
|
| 139 |
+
"model.language_model.layers.16.self_attn.o_proj.weight",
|
| 140 |
+
"model.language_model.layers.16.self_attn.q_proj.weight",
|
| 141 |
+
"model.language_model.layers.16.self_attn.v_proj.weight",
|
| 142 |
+
"model.language_model.layers.17.mlp.down_proj.weight",
|
| 143 |
+
"model.language_model.layers.17.mlp.gate_up_proj.weight",
|
| 144 |
+
"model.language_model.layers.17.self_attn.k_proj.weight",
|
| 145 |
+
"model.language_model.layers.17.self_attn.o_proj.weight",
|
| 146 |
+
"model.language_model.layers.17.self_attn.q_proj.weight",
|
| 147 |
+
"model.language_model.layers.17.self_attn.v_proj.weight",
|
| 148 |
+
"model.language_model.layers.18.mlp.down_proj.weight",
|
| 149 |
+
"model.language_model.layers.18.mlp.gate_up_proj.weight",
|
| 150 |
+
"model.language_model.layers.18.self_attn.k_proj.weight",
|
| 151 |
+
"model.language_model.layers.18.self_attn.o_proj.weight",
|
| 152 |
+
"model.language_model.layers.18.self_attn.q_proj.weight",
|
| 153 |
+
"model.language_model.layers.18.self_attn.v_proj.weight",
|
| 154 |
+
"model.language_model.layers.19.mlp.down_proj.weight",
|
| 155 |
+
"model.language_model.layers.19.mlp.gate_up_proj.weight",
|
| 156 |
+
"model.language_model.layers.19.self_attn.k_proj.weight",
|
| 157 |
+
"model.language_model.layers.19.self_attn.o_proj.weight",
|
| 158 |
+
"model.language_model.layers.19.self_attn.q_proj.weight",
|
| 159 |
+
"model.language_model.layers.19.self_attn.v_proj.weight",
|
| 160 |
+
"model.language_model.layers.20.mlp.down_proj.weight",
|
| 161 |
+
"model.language_model.layers.20.mlp.gate_up_proj.weight",
|
| 162 |
+
"model.language_model.layers.20.self_attn.k_proj.weight",
|
| 163 |
+
"model.language_model.layers.20.self_attn.o_proj.weight",
|
| 164 |
+
"model.language_model.layers.20.self_attn.q_proj.weight",
|
| 165 |
+
"model.language_model.layers.20.self_attn.v_proj.weight",
|
| 166 |
+
"model.language_model.layers.21.mlp.down_proj.weight",
|
| 167 |
+
"model.language_model.layers.21.mlp.gate_up_proj.weight",
|
| 168 |
+
"model.language_model.layers.21.self_attn.k_proj.weight",
|
| 169 |
+
"model.language_model.layers.21.self_attn.o_proj.weight",
|
| 170 |
+
"model.language_model.layers.21.self_attn.q_proj.weight",
|
| 171 |
+
"model.language_model.layers.21.self_attn.v_proj.weight",
|
| 172 |
+
"model.language_model.layers.22.mlp.down_proj.weight",
|
| 173 |
+
"model.language_model.layers.22.mlp.gate_up_proj.weight",
|
| 174 |
+
"model.language_model.layers.22.self_attn.k_proj.weight",
|
| 175 |
+
"model.language_model.layers.22.self_attn.o_proj.weight",
|
| 176 |
+
"model.language_model.layers.22.self_attn.q_proj.weight",
|
| 177 |
+
"model.language_model.layers.22.self_attn.v_proj.weight",
|
| 178 |
+
"model.language_model.layers.23.mlp.down_proj.weight",
|
| 179 |
+
"model.language_model.layers.23.mlp.gate_up_proj.weight",
|
| 180 |
+
"model.language_model.layers.23.self_attn.k_proj.weight",
|
| 181 |
+
"model.language_model.layers.23.self_attn.o_proj.weight",
|
| 182 |
+
"model.language_model.layers.23.self_attn.q_proj.weight",
|
| 183 |
+
"model.language_model.layers.23.self_attn.v_proj.weight",
|
| 184 |
+
"model.language_model.layers.24.mlp.down_proj.weight",
|
| 185 |
+
"model.language_model.layers.24.mlp.gate_up_proj.weight",
|
| 186 |
+
"model.language_model.layers.24.self_attn.k_proj.weight",
|
| 187 |
+
"model.language_model.layers.24.self_attn.o_proj.weight",
|
| 188 |
+
"model.language_model.layers.24.self_attn.q_proj.weight",
|
| 189 |
+
"model.language_model.layers.24.self_attn.v_proj.weight",
|
| 190 |
+
"model.language_model.layers.25.mlp.down_proj.weight",
|
| 191 |
+
"model.language_model.layers.25.mlp.gate_up_proj.weight",
|
| 192 |
+
"model.language_model.layers.25.self_attn.k_proj.weight",
|
| 193 |
+
"model.language_model.layers.25.self_attn.o_proj.weight",
|
| 194 |
+
"model.language_model.layers.25.self_attn.q_proj.weight",
|
| 195 |
+
"model.language_model.layers.25.self_attn.v_proj.weight",
|
| 196 |
+
"model.language_model.layers.26.mlp.down_proj.weight",
|
| 197 |
+
"model.language_model.layers.26.mlp.gate_up_proj.weight",
|
| 198 |
+
"model.language_model.layers.26.self_attn.k_proj.weight",
|
| 199 |
+
"model.language_model.layers.26.self_attn.o_proj.weight",
|
| 200 |
+
"model.language_model.layers.26.self_attn.q_proj.weight",
|
| 201 |
+
"model.language_model.layers.26.self_attn.v_proj.weight",
|
| 202 |
+
"model.language_model.layers.27.mlp.down_proj.weight",
|
| 203 |
+
"model.language_model.layers.27.mlp.gate_up_proj.weight",
|
| 204 |
+
"model.language_model.layers.27.self_attn.k_proj.weight",
|
| 205 |
+
"model.language_model.layers.27.self_attn.o_proj.weight",
|
| 206 |
+
"model.language_model.layers.27.self_attn.q_proj.weight",
|
| 207 |
+
"model.language_model.layers.27.self_attn.v_proj.weight",
|
| 208 |
+
"model.language_model.layers.28.mlp.down_proj.weight",
|
| 209 |
+
"model.language_model.layers.28.mlp.gate_up_proj.weight",
|
| 210 |
+
"model.language_model.layers.28.self_attn.k_proj.weight",
|
| 211 |
+
"model.language_model.layers.28.self_attn.o_proj.weight",
|
| 212 |
+
"model.language_model.layers.28.self_attn.q_proj.weight",
|
| 213 |
+
"model.language_model.layers.29.mlp.down_proj.weight",
|
| 214 |
+
"model.language_model.layers.29.mlp.gate_up_proj.weight",
|
| 215 |
+
"model.language_model.layers.29.self_attn.k_proj.weight",
|
| 216 |
+
"model.language_model.layers.29.self_attn.o_proj.weight",
|
| 217 |
+
"model.language_model.layers.29.self_attn.q_proj.weight",
|
| 218 |
+
"model.language_model.layers.30.mlp.down_proj.weight",
|
| 219 |
+
"model.language_model.layers.30.mlp.gate_up_proj.weight",
|
| 220 |
+
"model.language_model.layers.30.self_attn.k_proj.weight",
|
| 221 |
+
"model.language_model.layers.30.self_attn.o_proj.weight",
|
| 222 |
+
"model.language_model.layers.30.self_attn.q_proj.weight",
|
| 223 |
+
"model.language_model.layers.30.self_attn.v_proj.weight",
|
| 224 |
+
"model.language_model.layers.31.mlp.down_proj.weight",
|
| 225 |
+
"model.language_model.layers.31.mlp.gate_up_proj.weight",
|
| 226 |
+
"model.language_model.layers.31.self_attn.k_proj.weight",
|
| 227 |
+
"model.language_model.layers.31.self_attn.o_proj.weight",
|
| 228 |
+
"model.language_model.layers.31.self_attn.q_proj.weight",
|
| 229 |
+
"model.language_model.layers.31.self_attn.v_proj.weight",
|
| 230 |
+
"model.language_model.layers.32.mlp.down_proj.weight",
|
| 231 |
+
"model.language_model.layers.32.mlp.gate_up_proj.weight",
|
| 232 |
+
"model.language_model.layers.32.self_attn.k_proj.weight",
|
| 233 |
+
"model.language_model.layers.32.self_attn.o_proj.weight",
|
| 234 |
+
"model.language_model.layers.32.self_attn.q_proj.weight",
|
| 235 |
+
"model.language_model.layers.33.mlp.down_proj.weight",
|
| 236 |
+
"model.language_model.layers.33.mlp.gate_up_proj.weight",
|
| 237 |
+
"model.language_model.layers.33.self_attn.k_proj.weight",
|
| 238 |
+
"model.language_model.layers.33.self_attn.o_proj.weight",
|
| 239 |
+
"model.language_model.layers.33.self_attn.q_proj.weight",
|
| 240 |
+
"model.language_model.layers.34.mlp.down_proj.weight",
|
| 241 |
+
"model.language_model.layers.34.mlp.gate_up_proj.weight",
|
| 242 |
+
"model.language_model.layers.34.self_attn.k_proj.weight",
|
| 243 |
+
"model.language_model.layers.34.self_attn.o_proj.weight",
|
| 244 |
+
"model.language_model.layers.34.self_attn.q_proj.weight",
|
| 245 |
+
"model.language_model.layers.34.self_attn.v_proj.weight",
|
| 246 |
+
"model.language_model.layers.35.mlp.down_proj.weight",
|
| 247 |
+
"model.language_model.layers.35.mlp.gate_up_proj.weight",
|
| 248 |
+
"model.language_model.layers.35.self_attn.k_proj.weight",
|
| 249 |
+
"model.language_model.layers.35.self_attn.o_proj.weight",
|
| 250 |
+
"model.language_model.layers.35.self_attn.q_proj.weight",
|
| 251 |
+
"model.language_model.layers.35.self_attn.v_proj.weight",
|
| 252 |
+
"model.language_model.layers.36.mlp.down_proj.weight",
|
| 253 |
+
"model.language_model.layers.36.mlp.gate_up_proj.weight",
|
| 254 |
+
"model.language_model.layers.36.self_attn.k_proj.weight",
|
| 255 |
+
"model.language_model.layers.36.self_attn.o_proj.weight",
|
| 256 |
+
"model.language_model.layers.36.self_attn.q_proj.weight",
|
| 257 |
+
"model.language_model.layers.36.self_attn.v_proj.weight",
|
| 258 |
+
"model.language_model.layers.37.mlp.down_proj.weight",
|
| 259 |
+
"model.language_model.layers.37.mlp.gate_up_proj.weight",
|
| 260 |
+
"model.language_model.layers.37.self_attn.k_proj.weight",
|
| 261 |
+
"model.language_model.layers.37.self_attn.o_proj.weight",
|
| 262 |
+
"model.language_model.layers.37.self_attn.q_proj.weight",
|
| 263 |
+
"model.language_model.layers.37.self_attn.v_proj.weight",
|
| 264 |
+
"model.language_model.layers.38.mlp.down_proj.weight",
|
| 265 |
+
"model.language_model.layers.38.mlp.gate_up_proj.weight",
|
| 266 |
+
"model.language_model.layers.38.self_attn.k_proj.weight",
|
| 267 |
+
"model.language_model.layers.38.self_attn.o_proj.weight",
|
| 268 |
+
"model.language_model.layers.38.self_attn.q_proj.weight",
|
| 269 |
+
"model.language_model.layers.38.self_attn.v_proj.weight",
|
| 270 |
+
"model.language_model.layers.39.mlp.down_proj.weight",
|
| 271 |
+
"model.language_model.layers.39.mlp.gate_up_proj.weight",
|
| 272 |
+
"model.language_model.layers.39.self_attn.k_proj.weight",
|
| 273 |
+
"model.language_model.layers.39.self_attn.o_proj.weight",
|
| 274 |
+
"model.language_model.layers.39.self_attn.q_proj.weight",
|
| 275 |
+
"model.language_model.layers.39.self_attn.v_proj.weight",
|
| 276 |
+
"model.visual.blocks.0.attn.proj.weight",
|
| 277 |
+
"model.visual.blocks.0.mlp.fc1.weight",
|
| 278 |
+
"model.visual.blocks.1.attn.proj.weight",
|
| 279 |
+
"model.visual.blocks.2.attn.proj.weight",
|
| 280 |
+
"model.visual.blocks.2.mlp.fc1.weight",
|
| 281 |
+
"model.visual.blocks.3.attn.proj.weight",
|
| 282 |
+
"model.visual.blocks.3.mlp.fc1.weight",
|
| 283 |
+
"model.visual.blocks.4.attn.proj.weight",
|
| 284 |
+
"model.visual.blocks.4.attn.qkv.weight",
|
| 285 |
+
"model.visual.blocks.4.mlp.fc1.weight",
|
| 286 |
+
"model.visual.blocks.5.attn.proj.weight",
|
| 287 |
+
"model.visual.blocks.5.attn.qkv.weight",
|
| 288 |
+
"model.visual.blocks.5.mlp.fc1.weight",
|
| 289 |
+
"model.visual.blocks.6.attn.proj.weight",
|
| 290 |
+
"model.visual.blocks.6.attn.qkv.weight",
|
| 291 |
+
"model.visual.blocks.6.mlp.fc1.weight",
|
| 292 |
+
"model.visual.blocks.7.attn.proj.weight",
|
| 293 |
+
"model.visual.blocks.7.attn.qkv.weight",
|
| 294 |
+
"model.visual.blocks.7.mlp.fc1.weight",
|
| 295 |
+
"model.visual.blocks.8.attn.proj.weight",
|
| 296 |
+
"model.visual.blocks.8.attn.qkv.weight",
|
| 297 |
+
"model.visual.blocks.8.mlp.fc1.weight",
|
| 298 |
+
"model.visual.blocks.9.attn.proj.weight",
|
| 299 |
+
"model.visual.blocks.9.attn.qkv.weight",
|
| 300 |
+
"model.visual.blocks.9.mlp.fc1.weight",
|
| 301 |
+
"model.visual.blocks.10.attn.proj.weight",
|
| 302 |
+
"model.visual.blocks.10.attn.qkv.weight",
|
| 303 |
+
"model.visual.blocks.10.mlp.fc1.weight",
|
| 304 |
+
"model.visual.blocks.10.mlp.fc2.weight",
|
| 305 |
+
"model.visual.blocks.11.attn.proj.weight",
|
| 306 |
+
"model.visual.blocks.11.attn.qkv.weight",
|
| 307 |
+
"model.visual.blocks.11.mlp.fc1.weight",
|
| 308 |
+
"model.visual.blocks.11.mlp.fc2.weight",
|
| 309 |
+
"model.visual.blocks.12.attn.proj.weight",
|
| 310 |
+
"model.visual.blocks.12.attn.qkv.weight",
|
| 311 |
+
"model.visual.blocks.12.mlp.fc1.weight",
|
| 312 |
+
"model.visual.blocks.12.mlp.fc2.weight",
|
| 313 |
+
"model.visual.blocks.13.attn.proj.weight",
|
| 314 |
+
"model.visual.blocks.13.attn.qkv.weight",
|
| 315 |
+
"model.visual.blocks.13.mlp.fc1.weight",
|
| 316 |
+
"model.visual.blocks.13.mlp.fc2.weight",
|
| 317 |
+
"model.visual.blocks.14.attn.proj.weight",
|
| 318 |
+
"model.visual.blocks.14.attn.qkv.weight",
|
| 319 |
+
"model.visual.blocks.14.mlp.fc1.weight",
|
| 320 |
+
"model.visual.blocks.14.mlp.fc2.weight",
|
| 321 |
+
"model.visual.blocks.15.attn.proj.weight",
|
| 322 |
+
"model.visual.blocks.15.attn.qkv.weight",
|
| 323 |
+
"model.visual.blocks.15.mlp.fc1.weight",
|
| 324 |
+
"model.visual.blocks.15.mlp.fc2.weight",
|
| 325 |
+
"model.visual.blocks.16.attn.proj.weight",
|
| 326 |
+
"model.visual.blocks.16.attn.qkv.weight",
|
| 327 |
+
"model.visual.blocks.16.mlp.fc1.weight",
|
| 328 |
+
"model.visual.blocks.16.mlp.fc2.weight",
|
| 329 |
+
"model.visual.blocks.17.attn.proj.weight",
|
| 330 |
+
"model.visual.blocks.17.attn.qkv.weight",
|
| 331 |
+
"model.visual.blocks.17.mlp.fc1.weight",
|
| 332 |
+
"model.visual.blocks.17.mlp.fc2.weight",
|
| 333 |
+
"model.visual.blocks.18.attn.proj.weight",
|
| 334 |
+
"model.visual.blocks.18.attn.qkv.weight",
|
| 335 |
+
"model.visual.blocks.18.mlp.fc1.weight",
|
| 336 |
+
"model.visual.blocks.18.mlp.fc2.weight",
|
| 337 |
+
"model.visual.blocks.19.attn.proj.weight",
|
| 338 |
+
"model.visual.blocks.19.attn.qkv.weight",
|
| 339 |
+
"model.visual.blocks.19.mlp.fc1.weight",
|
| 340 |
+
"model.visual.blocks.19.mlp.fc2.weight",
|
| 341 |
+
"model.visual.blocks.20.attn.proj.weight",
|
| 342 |
+
"model.visual.blocks.20.attn.qkv.weight",
|
| 343 |
+
"model.visual.blocks.20.mlp.fc1.weight",
|
| 344 |
+
"model.visual.blocks.20.mlp.fc2.weight",
|
| 345 |
+
"model.visual.blocks.21.attn.proj.weight",
|
| 346 |
+
"model.visual.blocks.21.attn.qkv.weight",
|
| 347 |
+
"model.visual.blocks.21.mlp.fc1.weight",
|
| 348 |
+
"model.visual.blocks.21.mlp.fc2.weight",
|
| 349 |
+
"model.visual.blocks.22.attn.proj.weight",
|
| 350 |
+
"model.visual.blocks.22.attn.qkv.weight",
|
| 351 |
+
"model.visual.blocks.22.mlp.fc1.weight",
|
| 352 |
+
"model.visual.blocks.22.mlp.fc2.weight",
|
| 353 |
+
"model.visual.blocks.23.attn.proj.weight",
|
| 354 |
+
"model.visual.blocks.23.attn.qkv.weight",
|
| 355 |
+
"model.visual.blocks.23.mlp.fc1.weight",
|
| 356 |
+
"model.visual.blocks.23.mlp.fc2.weight",
|
| 357 |
+
"model.visual.blocks.24.attn.proj.weight",
|
| 358 |
+
"model.visual.blocks.24.attn.qkv.weight",
|
| 359 |
+
"model.visual.blocks.24.mlp.fc1.weight",
|
| 360 |
+
"model.visual.blocks.24.mlp.fc2.weight",
|
| 361 |
+
"model.visual.blocks.25.attn.proj.weight",
|
| 362 |
+
"model.visual.blocks.25.attn.qkv.weight",
|
| 363 |
+
"model.visual.blocks.25.mlp.fc1.weight",
|
| 364 |
+
"model.visual.blocks.25.mlp.fc2.weight",
|
| 365 |
+
"model.visual.blocks.26.attn.proj.weight",
|
| 366 |
+
"model.visual.blocks.26.attn.qkv.weight",
|
| 367 |
+
"model.visual.blocks.26.mlp.fc1.weight",
|
| 368 |
+
"model.visual.blocks.26.mlp.fc2.weight",
|
| 369 |
+
"model.visual.blocks.27.attn.proj.weight",
|
| 370 |
+
"model.visual.blocks.27.attn.qkv.weight",
|
| 371 |
+
"model.visual.blocks.27.mlp.fc1.weight",
|
| 372 |
+
"model.visual.blocks.27.mlp.fc2.weight",
|
| 373 |
+
"model.visual.blocks.28.attn.proj.weight",
|
| 374 |
+
"model.visual.blocks.28.attn.qkv.weight",
|
| 375 |
+
"model.visual.blocks.28.mlp.fc1.weight",
|
| 376 |
+
"model.visual.blocks.28.mlp.fc2.weight",
|
| 377 |
+
"model.visual.blocks.29.attn.proj.weight",
|
| 378 |
+
"model.visual.blocks.29.attn.qkv.weight",
|
| 379 |
+
"model.visual.blocks.29.mlp.fc1.weight",
|
| 380 |
+
"model.visual.blocks.29.mlp.fc2.weight",
|
| 381 |
+
"model.visual.blocks.30.attn.proj.weight",
|
| 382 |
+
"model.visual.blocks.30.attn.qkv.weight",
|
| 383 |
+
"model.visual.blocks.30.mlp.fc1.weight",
|
| 384 |
+
"model.visual.blocks.30.mlp.fc2.weight",
|
| 385 |
+
"model.visual.blocks.31.attn.proj.weight",
|
| 386 |
+
"model.visual.blocks.31.attn.qkv.weight",
|
| 387 |
+
"model.visual.blocks.31.mlp.fc1.weight",
|
| 388 |
+
"model.visual.blocks.31.mlp.fc2.weight",
|
| 389 |
+
"model.visual.blocks.32.attn.proj.weight",
|
| 390 |
+
"model.visual.blocks.32.attn.qkv.weight",
|
| 391 |
+
"model.visual.blocks.32.mlp.fc1.weight",
|
| 392 |
+
"model.visual.blocks.32.mlp.fc2.weight",
|
| 393 |
+
"model.visual.blocks.33.attn.proj.weight",
|
| 394 |
+
"model.visual.blocks.33.attn.qkv.weight",
|
| 395 |
+
"model.visual.blocks.33.mlp.fc1.weight",
|
| 396 |
+
"model.visual.blocks.33.mlp.fc2.weight",
|
| 397 |
+
"model.visual.blocks.34.attn.proj.weight",
|
| 398 |
+
"model.visual.blocks.34.attn.qkv.weight",
|
| 399 |
+
"model.visual.blocks.34.mlp.fc1.weight",
|
| 400 |
+
"model.visual.blocks.34.mlp.fc2.weight",
|
| 401 |
+
"model.visual.blocks.35.attn.proj.weight",
|
| 402 |
+
"model.visual.blocks.35.attn.qkv.weight",
|
| 403 |
+
"model.visual.blocks.35.mlp.fc1.weight",
|
| 404 |
+
"model.visual.blocks.35.mlp.fc2.weight",
|
| 405 |
+
"model.visual.blocks.36.attn.proj.weight",
|
| 406 |
+
"model.visual.blocks.36.attn.qkv.weight",
|
| 407 |
+
"model.visual.blocks.36.mlp.fc1.weight",
|
| 408 |
+
"model.visual.blocks.36.mlp.fc2.weight",
|
| 409 |
+
"model.visual.blocks.37.attn.proj.weight",
|
| 410 |
+
"model.visual.blocks.37.attn.qkv.weight",
|
| 411 |
+
"model.visual.blocks.37.mlp.fc1.weight",
|
| 412 |
+
"model.visual.blocks.37.mlp.fc2.weight",
|
| 413 |
+
"model.visual.blocks.38.attn.proj.weight",
|
| 414 |
+
"model.visual.blocks.38.attn.qkv.weight",
|
| 415 |
+
"model.visual.blocks.38.mlp.fc1.weight",
|
| 416 |
+
"model.visual.blocks.38.mlp.fc2.weight",
|
| 417 |
+
"model.visual.blocks.39.attn.proj.weight",
|
| 418 |
+
"model.visual.blocks.39.attn.qkv.weight",
|
| 419 |
+
"model.visual.blocks.39.mlp.fc1.weight",
|
| 420 |
+
"model.visual.blocks.39.mlp.fc2.weight"
|
| 421 |
+
]
|
| 422 |
+
},
|
| 423 |
+
"modules_to_not_convert": [
|
| 424 |
+
"vqmodel",
|
| 425 |
+
"lm_head",
|
| 426 |
+
"embeddings",
|
| 427 |
+
"embed_tokens",
|
| 428 |
+
"patch_embed"
|
| 429 |
+
],
|
| 430 |
+
"non_blocking": false,
|
| 431 |
+
"quant_conv": false,
|
| 432 |
+
"quant_method": "sdnq",
|
| 433 |
+
"quantization_device": null,
|
| 434 |
+
"quantized_matmul_dtype": null,
|
| 435 |
+
"return_device": null,
|
| 436 |
+
"sdnq_version": "0.1.4",
|
| 437 |
+
"svd_rank": 32,
|
| 438 |
+
"svd_steps": 8,
|
| 439 |
+
"use_dynamic_quantization": true,
|
| 440 |
+
"use_grad_ckpt": true,
|
| 441 |
+
"use_quantized_matmul": false,
|
| 442 |
+
"use_quantized_matmul_conv": false,
|
| 443 |
+
"use_static_quantization": true,
|
| 444 |
+
"use_stochastic_rounding": false,
|
| 445 |
+
"use_svd": false,
|
| 446 |
+
"weights_dtype": "uint4"
|
| 447 |
+
},
|
| 448 |
+
"text_config": {
|
| 449 |
+
"attention_bias": true,
|
| 450 |
+
"attention_dropout": 0.0,
|
| 451 |
+
"dtype": "bfloat16",
|
| 452 |
+
"eos_token_id": 16385,
|
| 453 |
+
"hidden_act": "silu",
|
| 454 |
+
"hidden_size": 4096,
|
| 455 |
+
"initializer_range": 0.02,
|
| 456 |
+
"intermediate_size": 13696,
|
| 457 |
+
"max_position_embeddings": 131072,
|
| 458 |
+
"model_type": "glm_image_text",
|
| 459 |
+
"num_attention_heads": 32,
|
| 460 |
+
"num_hidden_layers": 40,
|
| 461 |
+
"num_key_value_heads": 2,
|
| 462 |
+
"pad_token_id": 167841,
|
| 463 |
+
"rms_norm_eps": 1e-05,
|
| 464 |
+
"rope_parameters": {
|
| 465 |
+
"mrope_section": [
|
| 466 |
+
8,
|
| 467 |
+
12,
|
| 468 |
+
12
|
| 469 |
+
],
|
| 470 |
+
"partial_rotary_factor": 0.5,
|
| 471 |
+
"rope_theta": 10000,
|
| 472 |
+
"rope_type": "default"
|
| 473 |
+
},
|
| 474 |
+
"use_cache": true,
|
| 475 |
+
"vision_vocab_size": 16512,
|
| 476 |
+
"vocab_size": 168064
|
| 477 |
+
},
|
| 478 |
+
"transformers_version": "5.0.0.dev0",
|
| 479 |
+
"vision_config": {
|
| 480 |
+
"attention_bias": true,
|
| 481 |
+
"attention_dropout": 0.0,
|
| 482 |
+
"depth": 40,
|
| 483 |
+
"dtype": "bfloat16",
|
| 484 |
+
"hidden_act": "gelu",
|
| 485 |
+
"hidden_size": 1536,
|
| 486 |
+
"image_size": 2048,
|
| 487 |
+
"in_channels": 3,
|
| 488 |
+
"initializer_range": 0.02,
|
| 489 |
+
"intermediate_size": 6144,
|
| 490 |
+
"layer_norm_eps": 1e-06,
|
| 491 |
+
"model_type": "glm_image_vision",
|
| 492 |
+
"num_heads": 16,
|
| 493 |
+
"patch_size": 16,
|
| 494 |
+
"spatial_merge_size": 1
|
| 495 |
+
},
|
| 496 |
+
"vq_config": {
|
| 497 |
+
"dtype": "bfloat16",
|
| 498 |
+
"embed_dim": 2048,
|
| 499 |
+
"in_channels": 3,
|
| 500 |
+
"initializer_range": 0.02,
|
| 501 |
+
"latent_channels": 1536,
|
| 502 |
+
"model_type": "glm_image_vqmodel",
|
| 503 |
+
"num_embeddings": 16384
|
| 504 |
+
}
|
| 505 |
+
}
|
vision_language_encoder/generation_config.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": 16385,
|
| 5 |
+
"pad_token_id": 167855,
|
| 6 |
+
"temperature": 0.9,
|
| 7 |
+
"top_k": 16512,
|
| 8 |
+
"top_p": 0.75,
|
| 9 |
+
"transformers_version": "5.0.0.dev0"
|
| 10 |
+
}
|
vision_language_encoder/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:77a666a690adf0c16670b9613707f331bc2a9a4ecf4484a63c0c1dd6399defae
|
| 3 |
+
size 4990261848
|
vision_language_encoder/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2b1d6be8219eab4d13bb7c8a38f4d247400cb34bf11831ae482b10bebbc9fe54
|
| 3 |
+
size 1894171528
|
vision_language_encoder/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
vision_language_encoder/quantization_config.json
ADDED
|
@@ -0,0 +1,438 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_skip_keys": false,
|
| 3 |
+
"dequantize_fp32": false,
|
| 4 |
+
"dynamic_loss_threshold": 0.01,
|
| 5 |
+
"group_size": 0,
|
| 6 |
+
"is_integer": true,
|
| 7 |
+
"is_training": false,
|
| 8 |
+
"modules_dtype_dict": {
|
| 9 |
+
"int5": [
|
| 10 |
+
"model.language_model.layers.28.self_attn.v_proj.weight",
|
| 11 |
+
"model.language_model.layers.29.self_attn.v_proj.weight",
|
| 12 |
+
"model.language_model.layers.32.self_attn.v_proj.weight",
|
| 13 |
+
"model.language_model.layers.33.self_attn.v_proj.weight",
|
| 14 |
+
"model.visual.blocks.0.attn.qkv.weight",
|
| 15 |
+
"model.visual.blocks.0.mlp.fc2.weight",
|
| 16 |
+
"model.visual.blocks.1.attn.qkv.weight",
|
| 17 |
+
"model.visual.blocks.1.mlp.fc1.weight",
|
| 18 |
+
"model.visual.blocks.1.mlp.fc2.weight",
|
| 19 |
+
"model.visual.blocks.2.attn.qkv.weight",
|
| 20 |
+
"model.visual.blocks.2.mlp.fc2.weight",
|
| 21 |
+
"model.visual.blocks.3.attn.qkv.weight",
|
| 22 |
+
"model.visual.blocks.3.mlp.fc2.weight",
|
| 23 |
+
"model.visual.blocks.4.mlp.fc2.weight",
|
| 24 |
+
"model.visual.blocks.5.mlp.fc2.weight",
|
| 25 |
+
"model.visual.blocks.6.mlp.fc2.weight",
|
| 26 |
+
"model.visual.blocks.7.mlp.fc2.weight",
|
| 27 |
+
"model.visual.blocks.8.mlp.fc2.weight",
|
| 28 |
+
"model.visual.blocks.9.mlp.fc2.weight"
|
| 29 |
+
],
|
| 30 |
+
"uint4": [
|
| 31 |
+
"model.language_model.layers.0.mlp.down_proj.weight",
|
| 32 |
+
"model.language_model.layers.0.mlp.gate_up_proj.weight",
|
| 33 |
+
"model.language_model.layers.0.self_attn.k_proj.weight",
|
| 34 |
+
"model.language_model.layers.0.self_attn.o_proj.weight",
|
| 35 |
+
"model.language_model.layers.0.self_attn.q_proj.weight",
|
| 36 |
+
"model.language_model.layers.0.self_attn.v_proj.weight",
|
| 37 |
+
"model.language_model.layers.1.mlp.down_proj.weight",
|
| 38 |
+
"model.language_model.layers.1.mlp.gate_up_proj.weight",
|
| 39 |
+
"model.language_model.layers.1.self_attn.k_proj.weight",
|
| 40 |
+
"model.language_model.layers.1.self_attn.o_proj.weight",
|
| 41 |
+
"model.language_model.layers.1.self_attn.q_proj.weight",
|
| 42 |
+
"model.language_model.layers.1.self_attn.v_proj.weight",
|
| 43 |
+
"model.language_model.layers.2.mlp.down_proj.weight",
|
| 44 |
+
"model.language_model.layers.2.mlp.gate_up_proj.weight",
|
| 45 |
+
"model.language_model.layers.2.self_attn.k_proj.weight",
|
| 46 |
+
"model.language_model.layers.2.self_attn.o_proj.weight",
|
| 47 |
+
"model.language_model.layers.2.self_attn.q_proj.weight",
|
| 48 |
+
"model.language_model.layers.2.self_attn.v_proj.weight",
|
| 49 |
+
"model.language_model.layers.3.mlp.down_proj.weight",
|
| 50 |
+
"model.language_model.layers.3.mlp.gate_up_proj.weight",
|
| 51 |
+
"model.language_model.layers.3.self_attn.k_proj.weight",
|
| 52 |
+
"model.language_model.layers.3.self_attn.o_proj.weight",
|
| 53 |
+
"model.language_model.layers.3.self_attn.q_proj.weight",
|
| 54 |
+
"model.language_model.layers.3.self_attn.v_proj.weight",
|
| 55 |
+
"model.language_model.layers.4.mlp.down_proj.weight",
|
| 56 |
+
"model.language_model.layers.4.mlp.gate_up_proj.weight",
|
| 57 |
+
"model.language_model.layers.4.self_attn.k_proj.weight",
|
| 58 |
+
"model.language_model.layers.4.self_attn.o_proj.weight",
|
| 59 |
+
"model.language_model.layers.4.self_attn.q_proj.weight",
|
| 60 |
+
"model.language_model.layers.4.self_attn.v_proj.weight",
|
| 61 |
+
"model.language_model.layers.5.mlp.down_proj.weight",
|
| 62 |
+
"model.language_model.layers.5.mlp.gate_up_proj.weight",
|
| 63 |
+
"model.language_model.layers.5.self_attn.k_proj.weight",
|
| 64 |
+
"model.language_model.layers.5.self_attn.o_proj.weight",
|
| 65 |
+
"model.language_model.layers.5.self_attn.q_proj.weight",
|
| 66 |
+
"model.language_model.layers.5.self_attn.v_proj.weight",
|
| 67 |
+
"model.language_model.layers.6.mlp.down_proj.weight",
|
| 68 |
+
"model.language_model.layers.6.mlp.gate_up_proj.weight",
|
| 69 |
+
"model.language_model.layers.6.self_attn.k_proj.weight",
|
| 70 |
+
"model.language_model.layers.6.self_attn.o_proj.weight",
|
| 71 |
+
"model.language_model.layers.6.self_attn.q_proj.weight",
|
| 72 |
+
"model.language_model.layers.6.self_attn.v_proj.weight",
|
| 73 |
+
"model.language_model.layers.7.mlp.down_proj.weight",
|
| 74 |
+
"model.language_model.layers.7.mlp.gate_up_proj.weight",
|
| 75 |
+
"model.language_model.layers.7.self_attn.k_proj.weight",
|
| 76 |
+
"model.language_model.layers.7.self_attn.o_proj.weight",
|
| 77 |
+
"model.language_model.layers.7.self_attn.q_proj.weight",
|
| 78 |
+
"model.language_model.layers.7.self_attn.v_proj.weight",
|
| 79 |
+
"model.language_model.layers.8.mlp.down_proj.weight",
|
| 80 |
+
"model.language_model.layers.8.mlp.gate_up_proj.weight",
|
| 81 |
+
"model.language_model.layers.8.self_attn.k_proj.weight",
|
| 82 |
+
"model.language_model.layers.8.self_attn.o_proj.weight",
|
| 83 |
+
"model.language_model.layers.8.self_attn.q_proj.weight",
|
| 84 |
+
"model.language_model.layers.8.self_attn.v_proj.weight",
|
| 85 |
+
"model.language_model.layers.9.mlp.down_proj.weight",
|
| 86 |
+
"model.language_model.layers.9.mlp.gate_up_proj.weight",
|
| 87 |
+
"model.language_model.layers.9.self_attn.k_proj.weight",
|
| 88 |
+
"model.language_model.layers.9.self_attn.o_proj.weight",
|
| 89 |
+
"model.language_model.layers.9.self_attn.q_proj.weight",
|
| 90 |
+
"model.language_model.layers.9.self_attn.v_proj.weight",
|
| 91 |
+
"model.language_model.layers.10.mlp.down_proj.weight",
|
| 92 |
+
"model.language_model.layers.10.mlp.gate_up_proj.weight",
|
| 93 |
+
"model.language_model.layers.10.self_attn.k_proj.weight",
|
| 94 |
+
"model.language_model.layers.10.self_attn.o_proj.weight",
|
| 95 |
+
"model.language_model.layers.10.self_attn.q_proj.weight",
|
| 96 |
+
"model.language_model.layers.10.self_attn.v_proj.weight",
|
| 97 |
+
"model.language_model.layers.11.mlp.down_proj.weight",
|
| 98 |
+
"model.language_model.layers.11.mlp.gate_up_proj.weight",
|
| 99 |
+
"model.language_model.layers.11.self_attn.k_proj.weight",
|
| 100 |
+
"model.language_model.layers.11.self_attn.o_proj.weight",
|
| 101 |
+
"model.language_model.layers.11.self_attn.q_proj.weight",
|
| 102 |
+
"model.language_model.layers.11.self_attn.v_proj.weight",
|
| 103 |
+
"model.language_model.layers.12.mlp.down_proj.weight",
|
| 104 |
+
"model.language_model.layers.12.mlp.gate_up_proj.weight",
|
| 105 |
+
"model.language_model.layers.12.self_attn.k_proj.weight",
|
| 106 |
+
"model.language_model.layers.12.self_attn.o_proj.weight",
|
| 107 |
+
"model.language_model.layers.12.self_attn.q_proj.weight",
|
| 108 |
+
"model.language_model.layers.12.self_attn.v_proj.weight",
|
| 109 |
+
"model.language_model.layers.13.mlp.down_proj.weight",
|
| 110 |
+
"model.language_model.layers.13.mlp.gate_up_proj.weight",
|
| 111 |
+
"model.language_model.layers.13.self_attn.k_proj.weight",
|
| 112 |
+
"model.language_model.layers.13.self_attn.o_proj.weight",
|
| 113 |
+
"model.language_model.layers.13.self_attn.q_proj.weight",
|
| 114 |
+
"model.language_model.layers.13.self_attn.v_proj.weight",
|
| 115 |
+
"model.language_model.layers.14.mlp.down_proj.weight",
|
| 116 |
+
"model.language_model.layers.14.mlp.gate_up_proj.weight",
|
| 117 |
+
"model.language_model.layers.14.self_attn.k_proj.weight",
|
| 118 |
+
"model.language_model.layers.14.self_attn.o_proj.weight",
|
| 119 |
+
"model.language_model.layers.14.self_attn.q_proj.weight",
|
| 120 |
+
"model.language_model.layers.14.self_attn.v_proj.weight",
|
| 121 |
+
"model.language_model.layers.15.mlp.down_proj.weight",
|
| 122 |
+
"model.language_model.layers.15.mlp.gate_up_proj.weight",
|
| 123 |
+
"model.language_model.layers.15.self_attn.k_proj.weight",
|
| 124 |
+
"model.language_model.layers.15.self_attn.o_proj.weight",
|
| 125 |
+
"model.language_model.layers.15.self_attn.q_proj.weight",
|
| 126 |
+
"model.language_model.layers.15.self_attn.v_proj.weight",
|
| 127 |
+
"model.language_model.layers.16.mlp.down_proj.weight",
|
| 128 |
+
"model.language_model.layers.16.mlp.gate_up_proj.weight",
|
| 129 |
+
"model.language_model.layers.16.self_attn.k_proj.weight",
|
| 130 |
+
"model.language_model.layers.16.self_attn.o_proj.weight",
|
| 131 |
+
"model.language_model.layers.16.self_attn.q_proj.weight",
|
| 132 |
+
"model.language_model.layers.16.self_attn.v_proj.weight",
|
| 133 |
+
"model.language_model.layers.17.mlp.down_proj.weight",
|
| 134 |
+
"model.language_model.layers.17.mlp.gate_up_proj.weight",
|
| 135 |
+
"model.language_model.layers.17.self_attn.k_proj.weight",
|
| 136 |
+
"model.language_model.layers.17.self_attn.o_proj.weight",
|
| 137 |
+
"model.language_model.layers.17.self_attn.q_proj.weight",
|
| 138 |
+
"model.language_model.layers.17.self_attn.v_proj.weight",
|
| 139 |
+
"model.language_model.layers.18.mlp.down_proj.weight",
|
| 140 |
+
"model.language_model.layers.18.mlp.gate_up_proj.weight",
|
| 141 |
+
"model.language_model.layers.18.self_attn.k_proj.weight",
|
| 142 |
+
"model.language_model.layers.18.self_attn.o_proj.weight",
|
| 143 |
+
"model.language_model.layers.18.self_attn.q_proj.weight",
|
| 144 |
+
"model.language_model.layers.18.self_attn.v_proj.weight",
|
| 145 |
+
"model.language_model.layers.19.mlp.down_proj.weight",
|
| 146 |
+
"model.language_model.layers.19.mlp.gate_up_proj.weight",
|
| 147 |
+
"model.language_model.layers.19.self_attn.k_proj.weight",
|
| 148 |
+
"model.language_model.layers.19.self_attn.o_proj.weight",
|
| 149 |
+
"model.language_model.layers.19.self_attn.q_proj.weight",
|
| 150 |
+
"model.language_model.layers.19.self_attn.v_proj.weight",
|
| 151 |
+
"model.language_model.layers.20.mlp.down_proj.weight",
|
| 152 |
+
"model.language_model.layers.20.mlp.gate_up_proj.weight",
|
| 153 |
+
"model.language_model.layers.20.self_attn.k_proj.weight",
|
| 154 |
+
"model.language_model.layers.20.self_attn.o_proj.weight",
|
| 155 |
+
"model.language_model.layers.20.self_attn.q_proj.weight",
|
| 156 |
+
"model.language_model.layers.20.self_attn.v_proj.weight",
|
| 157 |
+
"model.language_model.layers.21.mlp.down_proj.weight",
|
| 158 |
+
"model.language_model.layers.21.mlp.gate_up_proj.weight",
|
| 159 |
+
"model.language_model.layers.21.self_attn.k_proj.weight",
|
| 160 |
+
"model.language_model.layers.21.self_attn.o_proj.weight",
|
| 161 |
+
"model.language_model.layers.21.self_attn.q_proj.weight",
|
| 162 |
+
"model.language_model.layers.21.self_attn.v_proj.weight",
|
| 163 |
+
"model.language_model.layers.22.mlp.down_proj.weight",
|
| 164 |
+
"model.language_model.layers.22.mlp.gate_up_proj.weight",
|
| 165 |
+
"model.language_model.layers.22.self_attn.k_proj.weight",
|
| 166 |
+
"model.language_model.layers.22.self_attn.o_proj.weight",
|
| 167 |
+
"model.language_model.layers.22.self_attn.q_proj.weight",
|
| 168 |
+
"model.language_model.layers.22.self_attn.v_proj.weight",
|
| 169 |
+
"model.language_model.layers.23.mlp.down_proj.weight",
|
| 170 |
+
"model.language_model.layers.23.mlp.gate_up_proj.weight",
|
| 171 |
+
"model.language_model.layers.23.self_attn.k_proj.weight",
|
| 172 |
+
"model.language_model.layers.23.self_attn.o_proj.weight",
|
| 173 |
+
"model.language_model.layers.23.self_attn.q_proj.weight",
|
| 174 |
+
"model.language_model.layers.23.self_attn.v_proj.weight",
|
| 175 |
+
"model.language_model.layers.24.mlp.down_proj.weight",
|
| 176 |
+
"model.language_model.layers.24.mlp.gate_up_proj.weight",
|
| 177 |
+
"model.language_model.layers.24.self_attn.k_proj.weight",
|
| 178 |
+
"model.language_model.layers.24.self_attn.o_proj.weight",
|
| 179 |
+
"model.language_model.layers.24.self_attn.q_proj.weight",
|
| 180 |
+
"model.language_model.layers.24.self_attn.v_proj.weight",
|
| 181 |
+
"model.language_model.layers.25.mlp.down_proj.weight",
|
| 182 |
+
"model.language_model.layers.25.mlp.gate_up_proj.weight",
|
| 183 |
+
"model.language_model.layers.25.self_attn.k_proj.weight",
|
| 184 |
+
"model.language_model.layers.25.self_attn.o_proj.weight",
|
| 185 |
+
"model.language_model.layers.25.self_attn.q_proj.weight",
|
| 186 |
+
"model.language_model.layers.25.self_attn.v_proj.weight",
|
| 187 |
+
"model.language_model.layers.26.mlp.down_proj.weight",
|
| 188 |
+
"model.language_model.layers.26.mlp.gate_up_proj.weight",
|
| 189 |
+
"model.language_model.layers.26.self_attn.k_proj.weight",
|
| 190 |
+
"model.language_model.layers.26.self_attn.o_proj.weight",
|
| 191 |
+
"model.language_model.layers.26.self_attn.q_proj.weight",
|
| 192 |
+
"model.language_model.layers.26.self_attn.v_proj.weight",
|
| 193 |
+
"model.language_model.layers.27.mlp.down_proj.weight",
|
| 194 |
+
"model.language_model.layers.27.mlp.gate_up_proj.weight",
|
| 195 |
+
"model.language_model.layers.27.self_attn.k_proj.weight",
|
| 196 |
+
"model.language_model.layers.27.self_attn.o_proj.weight",
|
| 197 |
+
"model.language_model.layers.27.self_attn.q_proj.weight",
|
| 198 |
+
"model.language_model.layers.27.self_attn.v_proj.weight",
|
| 199 |
+
"model.language_model.layers.28.mlp.down_proj.weight",
|
| 200 |
+
"model.language_model.layers.28.mlp.gate_up_proj.weight",
|
| 201 |
+
"model.language_model.layers.28.self_attn.k_proj.weight",
|
| 202 |
+
"model.language_model.layers.28.self_attn.o_proj.weight",
|
| 203 |
+
"model.language_model.layers.28.self_attn.q_proj.weight",
|
| 204 |
+
"model.language_model.layers.29.mlp.down_proj.weight",
|
| 205 |
+
"model.language_model.layers.29.mlp.gate_up_proj.weight",
|
| 206 |
+
"model.language_model.layers.29.self_attn.k_proj.weight",
|
| 207 |
+
"model.language_model.layers.29.self_attn.o_proj.weight",
|
| 208 |
+
"model.language_model.layers.29.self_attn.q_proj.weight",
|
| 209 |
+
"model.language_model.layers.30.mlp.down_proj.weight",
|
| 210 |
+
"model.language_model.layers.30.mlp.gate_up_proj.weight",
|
| 211 |
+
"model.language_model.layers.30.self_attn.k_proj.weight",
|
| 212 |
+
"model.language_model.layers.30.self_attn.o_proj.weight",
|
| 213 |
+
"model.language_model.layers.30.self_attn.q_proj.weight",
|
| 214 |
+
"model.language_model.layers.30.self_attn.v_proj.weight",
|
| 215 |
+
"model.language_model.layers.31.mlp.down_proj.weight",
|
| 216 |
+
"model.language_model.layers.31.mlp.gate_up_proj.weight",
|
| 217 |
+
"model.language_model.layers.31.self_attn.k_proj.weight",
|
| 218 |
+
"model.language_model.layers.31.self_attn.o_proj.weight",
|
| 219 |
+
"model.language_model.layers.31.self_attn.q_proj.weight",
|
| 220 |
+
"model.language_model.layers.31.self_attn.v_proj.weight",
|
| 221 |
+
"model.language_model.layers.32.mlp.down_proj.weight",
|
| 222 |
+
"model.language_model.layers.32.mlp.gate_up_proj.weight",
|
| 223 |
+
"model.language_model.layers.32.self_attn.k_proj.weight",
|
| 224 |
+
"model.language_model.layers.32.self_attn.o_proj.weight",
|
| 225 |
+
"model.language_model.layers.32.self_attn.q_proj.weight",
|
| 226 |
+
"model.language_model.layers.33.mlp.down_proj.weight",
|
| 227 |
+
"model.language_model.layers.33.mlp.gate_up_proj.weight",
|
| 228 |
+
"model.language_model.layers.33.self_attn.k_proj.weight",
|
| 229 |
+
"model.language_model.layers.33.self_attn.o_proj.weight",
|
| 230 |
+
"model.language_model.layers.33.self_attn.q_proj.weight",
|
| 231 |
+
"model.language_model.layers.34.mlp.down_proj.weight",
|
| 232 |
+
"model.language_model.layers.34.mlp.gate_up_proj.weight",
|
| 233 |
+
"model.language_model.layers.34.self_attn.k_proj.weight",
|
| 234 |
+
"model.language_model.layers.34.self_attn.o_proj.weight",
|
| 235 |
+
"model.language_model.layers.34.self_attn.q_proj.weight",
|
| 236 |
+
"model.language_model.layers.34.self_attn.v_proj.weight",
|
| 237 |
+
"model.language_model.layers.35.mlp.down_proj.weight",
|
| 238 |
+
"model.language_model.layers.35.mlp.gate_up_proj.weight",
|
| 239 |
+
"model.language_model.layers.35.self_attn.k_proj.weight",
|
| 240 |
+
"model.language_model.layers.35.self_attn.o_proj.weight",
|
| 241 |
+
"model.language_model.layers.35.self_attn.q_proj.weight",
|
| 242 |
+
"model.language_model.layers.35.self_attn.v_proj.weight",
|
| 243 |
+
"model.language_model.layers.36.mlp.down_proj.weight",
|
| 244 |
+
"model.language_model.layers.36.mlp.gate_up_proj.weight",
|
| 245 |
+
"model.language_model.layers.36.self_attn.k_proj.weight",
|
| 246 |
+
"model.language_model.layers.36.self_attn.o_proj.weight",
|
| 247 |
+
"model.language_model.layers.36.self_attn.q_proj.weight",
|
| 248 |
+
"model.language_model.layers.36.self_attn.v_proj.weight",
|
| 249 |
+
"model.language_model.layers.37.mlp.down_proj.weight",
|
| 250 |
+
"model.language_model.layers.37.mlp.gate_up_proj.weight",
|
| 251 |
+
"model.language_model.layers.37.self_attn.k_proj.weight",
|
| 252 |
+
"model.language_model.layers.37.self_attn.o_proj.weight",
|
| 253 |
+
"model.language_model.layers.37.self_attn.q_proj.weight",
|
| 254 |
+
"model.language_model.layers.37.self_attn.v_proj.weight",
|
| 255 |
+
"model.language_model.layers.38.mlp.down_proj.weight",
|
| 256 |
+
"model.language_model.layers.38.mlp.gate_up_proj.weight",
|
| 257 |
+
"model.language_model.layers.38.self_attn.k_proj.weight",
|
| 258 |
+
"model.language_model.layers.38.self_attn.o_proj.weight",
|
| 259 |
+
"model.language_model.layers.38.self_attn.q_proj.weight",
|
| 260 |
+
"model.language_model.layers.38.self_attn.v_proj.weight",
|
| 261 |
+
"model.language_model.layers.39.mlp.down_proj.weight",
|
| 262 |
+
"model.language_model.layers.39.mlp.gate_up_proj.weight",
|
| 263 |
+
"model.language_model.layers.39.self_attn.k_proj.weight",
|
| 264 |
+
"model.language_model.layers.39.self_attn.o_proj.weight",
|
| 265 |
+
"model.language_model.layers.39.self_attn.q_proj.weight",
|
| 266 |
+
"model.language_model.layers.39.self_attn.v_proj.weight",
|
| 267 |
+
"model.visual.blocks.0.attn.proj.weight",
|
| 268 |
+
"model.visual.blocks.0.mlp.fc1.weight",
|
| 269 |
+
"model.visual.blocks.1.attn.proj.weight",
|
| 270 |
+
"model.visual.blocks.2.attn.proj.weight",
|
| 271 |
+
"model.visual.blocks.2.mlp.fc1.weight",
|
| 272 |
+
"model.visual.blocks.3.attn.proj.weight",
|
| 273 |
+
"model.visual.blocks.3.mlp.fc1.weight",
|
| 274 |
+
"model.visual.blocks.4.attn.proj.weight",
|
| 275 |
+
"model.visual.blocks.4.attn.qkv.weight",
|
| 276 |
+
"model.visual.blocks.4.mlp.fc1.weight",
|
| 277 |
+
"model.visual.blocks.5.attn.proj.weight",
|
| 278 |
+
"model.visual.blocks.5.attn.qkv.weight",
|
| 279 |
+
"model.visual.blocks.5.mlp.fc1.weight",
|
| 280 |
+
"model.visual.blocks.6.attn.proj.weight",
|
| 281 |
+
"model.visual.blocks.6.attn.qkv.weight",
|
| 282 |
+
"model.visual.blocks.6.mlp.fc1.weight",
|
| 283 |
+
"model.visual.blocks.7.attn.proj.weight",
|
| 284 |
+
"model.visual.blocks.7.attn.qkv.weight",
|
| 285 |
+
"model.visual.blocks.7.mlp.fc1.weight",
|
| 286 |
+
"model.visual.blocks.8.attn.proj.weight",
|
| 287 |
+
"model.visual.blocks.8.attn.qkv.weight",
|
| 288 |
+
"model.visual.blocks.8.mlp.fc1.weight",
|
| 289 |
+
"model.visual.blocks.9.attn.proj.weight",
|
| 290 |
+
"model.visual.blocks.9.attn.qkv.weight",
|
| 291 |
+
"model.visual.blocks.9.mlp.fc1.weight",
|
| 292 |
+
"model.visual.blocks.10.attn.proj.weight",
|
| 293 |
+
"model.visual.blocks.10.attn.qkv.weight",
|
| 294 |
+
"model.visual.blocks.10.mlp.fc1.weight",
|
| 295 |
+
"model.visual.blocks.10.mlp.fc2.weight",
|
| 296 |
+
"model.visual.blocks.11.attn.proj.weight",
|
| 297 |
+
"model.visual.blocks.11.attn.qkv.weight",
|
| 298 |
+
"model.visual.blocks.11.mlp.fc1.weight",
|
| 299 |
+
"model.visual.blocks.11.mlp.fc2.weight",
|
| 300 |
+
"model.visual.blocks.12.attn.proj.weight",
|
| 301 |
+
"model.visual.blocks.12.attn.qkv.weight",
|
| 302 |
+
"model.visual.blocks.12.mlp.fc1.weight",
|
| 303 |
+
"model.visual.blocks.12.mlp.fc2.weight",
|
| 304 |
+
"model.visual.blocks.13.attn.proj.weight",
|
| 305 |
+
"model.visual.blocks.13.attn.qkv.weight",
|
| 306 |
+
"model.visual.blocks.13.mlp.fc1.weight",
|
| 307 |
+
"model.visual.blocks.13.mlp.fc2.weight",
|
| 308 |
+
"model.visual.blocks.14.attn.proj.weight",
|
| 309 |
+
"model.visual.blocks.14.attn.qkv.weight",
|
| 310 |
+
"model.visual.blocks.14.mlp.fc1.weight",
|
| 311 |
+
"model.visual.blocks.14.mlp.fc2.weight",
|
| 312 |
+
"model.visual.blocks.15.attn.proj.weight",
|
| 313 |
+
"model.visual.blocks.15.attn.qkv.weight",
|
| 314 |
+
"model.visual.blocks.15.mlp.fc1.weight",
|
| 315 |
+
"model.visual.blocks.15.mlp.fc2.weight",
|
| 316 |
+
"model.visual.blocks.16.attn.proj.weight",
|
| 317 |
+
"model.visual.blocks.16.attn.qkv.weight",
|
| 318 |
+
"model.visual.blocks.16.mlp.fc1.weight",
|
| 319 |
+
"model.visual.blocks.16.mlp.fc2.weight",
|
| 320 |
+
"model.visual.blocks.17.attn.proj.weight",
|
| 321 |
+
"model.visual.blocks.17.attn.qkv.weight",
|
| 322 |
+
"model.visual.blocks.17.mlp.fc1.weight",
|
| 323 |
+
"model.visual.blocks.17.mlp.fc2.weight",
|
| 324 |
+
"model.visual.blocks.18.attn.proj.weight",
|
| 325 |
+
"model.visual.blocks.18.attn.qkv.weight",
|
| 326 |
+
"model.visual.blocks.18.mlp.fc1.weight",
|
| 327 |
+
"model.visual.blocks.18.mlp.fc2.weight",
|
| 328 |
+
"model.visual.blocks.19.attn.proj.weight",
|
| 329 |
+
"model.visual.blocks.19.attn.qkv.weight",
|
| 330 |
+
"model.visual.blocks.19.mlp.fc1.weight",
|
| 331 |
+
"model.visual.blocks.19.mlp.fc2.weight",
|
| 332 |
+
"model.visual.blocks.20.attn.proj.weight",
|
| 333 |
+
"model.visual.blocks.20.attn.qkv.weight",
|
| 334 |
+
"model.visual.blocks.20.mlp.fc1.weight",
|
| 335 |
+
"model.visual.blocks.20.mlp.fc2.weight",
|
| 336 |
+
"model.visual.blocks.21.attn.proj.weight",
|
| 337 |
+
"model.visual.blocks.21.attn.qkv.weight",
|
| 338 |
+
"model.visual.blocks.21.mlp.fc1.weight",
|
| 339 |
+
"model.visual.blocks.21.mlp.fc2.weight",
|
| 340 |
+
"model.visual.blocks.22.attn.proj.weight",
|
| 341 |
+
"model.visual.blocks.22.attn.qkv.weight",
|
| 342 |
+
"model.visual.blocks.22.mlp.fc1.weight",
|
| 343 |
+
"model.visual.blocks.22.mlp.fc2.weight",
|
| 344 |
+
"model.visual.blocks.23.attn.proj.weight",
|
| 345 |
+
"model.visual.blocks.23.attn.qkv.weight",
|
| 346 |
+
"model.visual.blocks.23.mlp.fc1.weight",
|
| 347 |
+
"model.visual.blocks.23.mlp.fc2.weight",
|
| 348 |
+
"model.visual.blocks.24.attn.proj.weight",
|
| 349 |
+
"model.visual.blocks.24.attn.qkv.weight",
|
| 350 |
+
"model.visual.blocks.24.mlp.fc1.weight",
|
| 351 |
+
"model.visual.blocks.24.mlp.fc2.weight",
|
| 352 |
+
"model.visual.blocks.25.attn.proj.weight",
|
| 353 |
+
"model.visual.blocks.25.attn.qkv.weight",
|
| 354 |
+
"model.visual.blocks.25.mlp.fc1.weight",
|
| 355 |
+
"model.visual.blocks.25.mlp.fc2.weight",
|
| 356 |
+
"model.visual.blocks.26.attn.proj.weight",
|
| 357 |
+
"model.visual.blocks.26.attn.qkv.weight",
|
| 358 |
+
"model.visual.blocks.26.mlp.fc1.weight",
|
| 359 |
+
"model.visual.blocks.26.mlp.fc2.weight",
|
| 360 |
+
"model.visual.blocks.27.attn.proj.weight",
|
| 361 |
+
"model.visual.blocks.27.attn.qkv.weight",
|
| 362 |
+
"model.visual.blocks.27.mlp.fc1.weight",
|
| 363 |
+
"model.visual.blocks.27.mlp.fc2.weight",
|
| 364 |
+
"model.visual.blocks.28.attn.proj.weight",
|
| 365 |
+
"model.visual.blocks.28.attn.qkv.weight",
|
| 366 |
+
"model.visual.blocks.28.mlp.fc1.weight",
|
| 367 |
+
"model.visual.blocks.28.mlp.fc2.weight",
|
| 368 |
+
"model.visual.blocks.29.attn.proj.weight",
|
| 369 |
+
"model.visual.blocks.29.attn.qkv.weight",
|
| 370 |
+
"model.visual.blocks.29.mlp.fc1.weight",
|
| 371 |
+
"model.visual.blocks.29.mlp.fc2.weight",
|
| 372 |
+
"model.visual.blocks.30.attn.proj.weight",
|
| 373 |
+
"model.visual.blocks.30.attn.qkv.weight",
|
| 374 |
+
"model.visual.blocks.30.mlp.fc1.weight",
|
| 375 |
+
"model.visual.blocks.30.mlp.fc2.weight",
|
| 376 |
+
"model.visual.blocks.31.attn.proj.weight",
|
| 377 |
+
"model.visual.blocks.31.attn.qkv.weight",
|
| 378 |
+
"model.visual.blocks.31.mlp.fc1.weight",
|
| 379 |
+
"model.visual.blocks.31.mlp.fc2.weight",
|
| 380 |
+
"model.visual.blocks.32.attn.proj.weight",
|
| 381 |
+
"model.visual.blocks.32.attn.qkv.weight",
|
| 382 |
+
"model.visual.blocks.32.mlp.fc1.weight",
|
| 383 |
+
"model.visual.blocks.32.mlp.fc2.weight",
|
| 384 |
+
"model.visual.blocks.33.attn.proj.weight",
|
| 385 |
+
"model.visual.blocks.33.attn.qkv.weight",
|
| 386 |
+
"model.visual.blocks.33.mlp.fc1.weight",
|
| 387 |
+
"model.visual.blocks.33.mlp.fc2.weight",
|
| 388 |
+
"model.visual.blocks.34.attn.proj.weight",
|
| 389 |
+
"model.visual.blocks.34.attn.qkv.weight",
|
| 390 |
+
"model.visual.blocks.34.mlp.fc1.weight",
|
| 391 |
+
"model.visual.blocks.34.mlp.fc2.weight",
|
| 392 |
+
"model.visual.blocks.35.attn.proj.weight",
|
| 393 |
+
"model.visual.blocks.35.attn.qkv.weight",
|
| 394 |
+
"model.visual.blocks.35.mlp.fc1.weight",
|
| 395 |
+
"model.visual.blocks.35.mlp.fc2.weight",
|
| 396 |
+
"model.visual.blocks.36.attn.proj.weight",
|
| 397 |
+
"model.visual.blocks.36.attn.qkv.weight",
|
| 398 |
+
"model.visual.blocks.36.mlp.fc1.weight",
|
| 399 |
+
"model.visual.blocks.36.mlp.fc2.weight",
|
| 400 |
+
"model.visual.blocks.37.attn.proj.weight",
|
| 401 |
+
"model.visual.blocks.37.attn.qkv.weight",
|
| 402 |
+
"model.visual.blocks.37.mlp.fc1.weight",
|
| 403 |
+
"model.visual.blocks.37.mlp.fc2.weight",
|
| 404 |
+
"model.visual.blocks.38.attn.proj.weight",
|
| 405 |
+
"model.visual.blocks.38.attn.qkv.weight",
|
| 406 |
+
"model.visual.blocks.38.mlp.fc1.weight",
|
| 407 |
+
"model.visual.blocks.38.mlp.fc2.weight",
|
| 408 |
+
"model.visual.blocks.39.attn.proj.weight",
|
| 409 |
+
"model.visual.blocks.39.attn.qkv.weight",
|
| 410 |
+
"model.visual.blocks.39.mlp.fc1.weight",
|
| 411 |
+
"model.visual.blocks.39.mlp.fc2.weight"
|
| 412 |
+
]
|
| 413 |
+
},
|
| 414 |
+
"modules_to_not_convert": [
|
| 415 |
+
"vqmodel",
|
| 416 |
+
"lm_head",
|
| 417 |
+
"embeddings",
|
| 418 |
+
"embed_tokens",
|
| 419 |
+
"patch_embed"
|
| 420 |
+
],
|
| 421 |
+
"non_blocking": false,
|
| 422 |
+
"quant_conv": false,
|
| 423 |
+
"quant_method": "sdnq",
|
| 424 |
+
"quantization_device": null,
|
| 425 |
+
"quantized_matmul_dtype": null,
|
| 426 |
+
"return_device": null,
|
| 427 |
+
"sdnq_version": "0.1.4",
|
| 428 |
+
"svd_rank": 32,
|
| 429 |
+
"svd_steps": 8,
|
| 430 |
+
"use_dynamic_quantization": true,
|
| 431 |
+
"use_grad_ckpt": true,
|
| 432 |
+
"use_quantized_matmul": false,
|
| 433 |
+
"use_quantized_matmul_conv": false,
|
| 434 |
+
"use_static_quantization": true,
|
| 435 |
+
"use_stochastic_rounding": false,
|
| 436 |
+
"use_svd": false,
|
| 437 |
+
"weights_dtype": "uint4"
|
| 438 |
+
}
|