qazimbhat1
commited on
Commit
•
611f17e
1
Parent(s):
de7ed9c
Add initial model files
Browse files- config.json +52 -0
- mm_projector.bin +3 -0
- trainer_state.json +0 -0
config.json
ADDED
@@ -0,0 +1,52 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "LLM360/CrystalChat",
|
3 |
+
"activation_function": "swiglu",
|
4 |
+
"architectures": [
|
5 |
+
"CrystalCoderLMHeadModel"
|
6 |
+
],
|
7 |
+
"attn_pdrop": 0.0,
|
8 |
+
"auto_map": {
|
9 |
+
"AutoConfig": "LLM360/CrystalChat--configuration_crystalcoder.CrystalCoderConfig",
|
10 |
+
"AutoModel": "LLM360/CrystalChat--modeling_crystalcoder.CrystalCoderModel",
|
11 |
+
"AutoModelForCausalLM": "LLM360/CrystalChat--modeling_crystalcoder.CrystalCoderLMHeadModel"
|
12 |
+
},
|
13 |
+
"bos_token_id": 1,
|
14 |
+
"embd_pdrop": 0.0,
|
15 |
+
"eos_token_id": 2,
|
16 |
+
"freeze_mm_mlp_adapter": false,
|
17 |
+
"image_aspect_ratio": "square",
|
18 |
+
"initializer_range": 0.02,
|
19 |
+
"layer_norm_epsilon": 1e-05,
|
20 |
+
"mm_hidden_size": 1024,
|
21 |
+
"mm_projector_lr": null,
|
22 |
+
"mm_projector_type": "mlp2x_gelu",
|
23 |
+
"mm_use_im_patch_token": false,
|
24 |
+
"mm_use_im_start_end": false,
|
25 |
+
"mm_vision_select_feature": "patch",
|
26 |
+
"mm_vision_select_layer": -2,
|
27 |
+
"mm_vision_tower": "openai/clip-vit-large-patch14-336",
|
28 |
+
"model_type": "llava_crystal",
|
29 |
+
"mup_embeddings_scale": 14.6,
|
30 |
+
"mup_output_alpha": 2.22,
|
31 |
+
"mup_scale_qk_dot_by_d": true,
|
32 |
+
"mup_width_scale": 0.0625,
|
33 |
+
"n_embd": 4096,
|
34 |
+
"n_head": 32,
|
35 |
+
"n_inner": 10922,
|
36 |
+
"n_layer": 32,
|
37 |
+
"n_positions": 2048,
|
38 |
+
"position_embedding_type": "rotary",
|
39 |
+
"reorder_and_upcast_attn": false,
|
40 |
+
"resid_pdrop": 0.0,
|
41 |
+
"rotary_dim": 32,
|
42 |
+
"scale_attn_by_inverse_layer_idx": false,
|
43 |
+
"scale_attn_weights": true,
|
44 |
+
"tokenizer_model_max_length": 2048,
|
45 |
+
"tokenizer_padding_side": "right",
|
46 |
+
"torch_dtype": "bfloat16",
|
47 |
+
"transformers_version": "4.31.0",
|
48 |
+
"tune_mm_mlp_adapter": true,
|
49 |
+
"use_cache": true,
|
50 |
+
"use_mm_proj": true,
|
51 |
+
"vocab_size": 32032
|
52 |
+
}
|
mm_projector.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:95e8ca7ca93dbfd840acc70329a471da5b11dfe4db14d03eb61aab184e6c413a
|
3 |
+
size 41961085
|
trainer_state.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|