Upload 7 files
Browse files- added_tokens.json +5 -0
- audio_tower.bin +3 -0
- config.json +46 -0
- generation_config.json +14 -0
- merges.txt +0 -0
- mm_projector_a.bin +3 -0
- pytorch_model.bin.index.json +0 -0
added_tokens.json
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<|endoftext|>": 151643,
|
| 3 |
+
"<|im_end|>": 151645,
|
| 4 |
+
"<|im_start|>": 151644
|
| 5 |
+
}
|
audio_tower.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:574619cf9d7dc2ad577e38d2b5548d3dcc28e05c01d0f1bdf2c1f348d60b3c84
|
| 3 |
+
size 181514532
|
config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "caoyuqin/AGAV-Rater",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"Videollama2Qwen2ForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"bos_token_id": 151643,
|
| 8 |
+
"eos_token_id": 151645,
|
| 9 |
+
"freeze_mm_mlp_adapter": false,
|
| 10 |
+
"hidden_act": "silu",
|
| 11 |
+
"hidden_size": 3584,
|
| 12 |
+
"hidden_size_a": 3584,
|
| 13 |
+
"image_aspect_ratio": "square",
|
| 14 |
+
"initializer_range": 0.02,
|
| 15 |
+
"intermediate_size": 18944,
|
| 16 |
+
"max_position_embeddings": 32768,
|
| 17 |
+
"max_window_layers": 28,
|
| 18 |
+
"mm_audio_tower": "work_dirs/AGAV-Rater/audio_tower.bin",
|
| 19 |
+
"mm_hidden_size": 1152,
|
| 20 |
+
"mm_hidden_size_a": 768,
|
| 21 |
+
"mm_projector_a_type": "mlp2x_gelu",
|
| 22 |
+
"mm_projector_lr": null,
|
| 23 |
+
"mm_projector_type": "stc_connector_v35",
|
| 24 |
+
"mm_vision_select_feature": "patch",
|
| 25 |
+
"mm_vision_select_layer": -2,
|
| 26 |
+
"mm_vision_tower": "google/siglip-so400m-patch14-384",
|
| 27 |
+
"model_type": "videollama2_qwen2",
|
| 28 |
+
"num_attention_heads": 28,
|
| 29 |
+
"num_frames": 8,
|
| 30 |
+
"num_hidden_layers": 28,
|
| 31 |
+
"num_key_value_heads": 4,
|
| 32 |
+
"rms_norm_eps": 1e-06,
|
| 33 |
+
"rope_theta": 1000000.0,
|
| 34 |
+
"sliding_window": 131072,
|
| 35 |
+
"tie_word_embeddings": false,
|
| 36 |
+
"tokenizer_model_max_length": 2048,
|
| 37 |
+
"tokenizer_padding_side": "right",
|
| 38 |
+
"torch_dtype": "bfloat16",
|
| 39 |
+
"transformers_version": "4.42.3",
|
| 40 |
+
"tune_mm_mlp_adapter": false,
|
| 41 |
+
"tune_mm_mlp_adapter_a": true,
|
| 42 |
+
"use_cache": true,
|
| 43 |
+
"use_mm_proj": true,
|
| 44 |
+
"use_sliding_window": false,
|
| 45 |
+
"vocab_size": 152064
|
| 46 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 151643,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
151645,
|
| 6 |
+
151643
|
| 7 |
+
],
|
| 8 |
+
"pad_token_id": 151643,
|
| 9 |
+
"repetition_penalty": 1.05,
|
| 10 |
+
"temperature": 0.7,
|
| 11 |
+
"top_k": 20,
|
| 12 |
+
"top_p": 0.8,
|
| 13 |
+
"transformers_version": "4.42.3"
|
| 14 |
+
}
|
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
mm_projector_a.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3f30404054365b4467d71237fad94f2a428c0e886a7fad4378348046290e841b
|
| 3 |
+
size 31211592
|
pytorch_model.bin.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|