Nagase-Kotono commited on
Commit
7c81040
1 Parent(s): f2a5637
checkpoint-2000/config.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "yanolja/EEVE-Korean-Instruct-2.8B-v1.0",
3
+ "architectures": [
4
+ "PhiForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "auto_map": {
8
+ "AutoConfig": "microsoft/phi-2--configuration_phi.PhiConfig",
9
+ "AutoModelForCausalLM": "microsoft/phi-2--modeling_phi.PhiForCausalLM"
10
+ },
11
+ "bos_token_id": 58940,
12
+ "embd_pdrop": 0.0,
13
+ "eos_token_id": 58943,
14
+ "freeze_mm_mlp_adapter": false,
15
+ "hidden_act": "gelu_new",
16
+ "hidden_size": 2560,
17
+ "image_aspect_ratio": "square",
18
+ "image_projector_type": "mlp2x_gelu",
19
+ "initializer_range": 0.02,
20
+ "intermediate_size": 10240,
21
+ "layer_norm_eps": 1e-05,
22
+ "max_position_embeddings": 2048,
23
+ "mm_hidden_size": 1024,
24
+ "mm_image_tower": "openai/clip-vit-large-patch14-336",
25
+ "mm_projector_lr": null,
26
+ "mm_use_im_patch_token": false,
27
+ "mm_use_im_start_end": false,
28
+ "mm_video_tower": null,
29
+ "mm_vision_select_feature": "patch",
30
+ "mm_vision_select_layer": -2,
31
+ "model_type": "llava_phi",
32
+ "num_attention_heads": 32,
33
+ "num_hidden_layers": 32,
34
+ "num_key_value_heads": 32,
35
+ "pad_token_id": 58944,
36
+ "partial_rotary_factor": 0.4,
37
+ "qk_layernorm": false,
38
+ "resid_pdrop": 0.1,
39
+ "rope_scaling": null,
40
+ "rope_theta": 10000.0,
41
+ "tie_word_embeddings": false,
42
+ "tokenizer_padding_side": "right",
43
+ "torch_dtype": "bfloat16",
44
+ "transformers_version": "4.37.0",
45
+ "tune_mm_mlp_adapter": true,
46
+ "use_cache": false,
47
+ "use_mm_proj": true,
48
+ "video_global_proj": false,
49
+ "video_projector_type": "linear",
50
+ "video_spatial_proj": false,
51
+ "video_temproal_proj": false,
52
+ "vocab_size": 58944
53
+ }
checkpoint-2000/mm_projector.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b68e2aafe71aa7fa3a2fc9f379485b5088761b096c7ed4acc7e613d1e7c7bf8a
3
+ size 18362045
config.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "yanolja/EEVE-Korean-Instruct-2.8B-v1.0",
3
+ "architectures": [
4
+ "PhiForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "auto_map": {
8
+ "AutoConfig": "microsoft/phi-2--configuration_phi.PhiConfig",
9
+ "AutoModelForCausalLM": "microsoft/phi-2--modeling_phi.PhiForCausalLM"
10
+ },
11
+ "bos_token_id": 58940,
12
+ "embd_pdrop": 0.0,
13
+ "eos_token_id": 58943,
14
+ "freeze_mm_mlp_adapter": false,
15
+ "hidden_act": "gelu_new",
16
+ "hidden_size": 2560,
17
+ "image_aspect_ratio": "square",
18
+ "image_projector_type": "mlp2x_gelu",
19
+ "initializer_range": 0.02,
20
+ "intermediate_size": 10240,
21
+ "layer_norm_eps": 1e-05,
22
+ "max_position_embeddings": 2048,
23
+ "mm_hidden_size": 1024,
24
+ "mm_image_tower": "openai/clip-vit-large-patch14-336",
25
+ "mm_projector_lr": null,
26
+ "mm_use_im_patch_token": false,
27
+ "mm_use_im_start_end": false,
28
+ "mm_video_tower": null,
29
+ "mm_vision_select_feature": "patch",
30
+ "mm_vision_select_layer": -2,
31
+ "model_type": "llava_phi",
32
+ "num_attention_heads": 32,
33
+ "num_hidden_layers": 32,
34
+ "num_key_value_heads": 32,
35
+ "pad_token_id": 58944,
36
+ "partial_rotary_factor": 0.4,
37
+ "qk_layernorm": false,
38
+ "resid_pdrop": 0.1,
39
+ "rope_scaling": null,
40
+ "rope_theta": 10000.0,
41
+ "tie_word_embeddings": false,
42
+ "tokenizer_padding_side": "right",
43
+ "torch_dtype": "bfloat16",
44
+ "transformers_version": "4.37.0",
45
+ "tune_mm_mlp_adapter": true,
46
+ "use_cache": true,
47
+ "use_mm_proj": true,
48
+ "video_global_proj": false,
49
+ "video_projector_type": "linear",
50
+ "video_spatial_proj": false,
51
+ "video_temproal_proj": false,
52
+ "vocab_size": 58944
53
+ }
mm_projector.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e68d73dd08520f9996d72a553f0dcfd9f5fa1066a7c88e64376496483b22f7e5
3
+ size 18362045
trainer_state.json ADDED
The diff for this file is too large to render. See raw diff