RussRobin commited on
Commit
0f2e84b
1 Parent(s): 150780b

Upload 3 files

Browse files
bunny-phi-2-pretrain/bunny-phi-2-pretrain_config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "BAAI/bunny-pretrain-phi-2-siglip",
3
+ "architectures": [
4
+ "PhiForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 50256,
8
+ "embd_pdrop": 0.0,
9
+ "eos_token_id": 50256,
10
+ "freeze_mm_mlp_adapter": false,
11
+ "hidden_act": "gelu_new",
12
+ "hidden_size": 2560,
13
+ "image_aspect_ratio": "square",
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 10240,
16
+ "layer_norm_eps": 1e-05,
17
+ "max_position_embeddings": 2048,
18
+ "mm_hidden_size": 1152,
19
+ "mm_projector_lr": null,
20
+ "mm_projector_type": "mlp2x_gelu",
21
+ "mm_vision_select_feature": "patch",
22
+ "mm_vision_tower": "google/siglip-so400m-patch14-384",
23
+ "model_type": "bunny-phi",
24
+ "num_attention_heads": 32,
25
+ "num_hidden_layers": 32,
26
+ "partial_rotary_factor": 0.4,
27
+ "pretraining_tp": 1,
28
+ "qk_layernorm": false,
29
+ "resid_pdrop": 0.1,
30
+ "rope_scaling": null,
31
+ "rope_theta": 10000.0,
32
+ "tie_word_embeddings": false,
33
+ "tokenizer_model_max_length": 2048,
34
+ "tokenizer_padding_side": "right",
35
+ "torch_dtype": "float16",
36
+ "transformers_version": "4.36.2",
37
+ "tune_mm_mlp_adapter": true,
38
+ "use_cache": true,
39
+ "use_mm_proj": true,
40
+ "vocab_size": 51200
41
+ }
bunny-phi-2-pretrain/bunny-phi-2-pretrain_trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
bunny-phi-2-pretrain/mm_projector.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:528be96721288621bf02bae6fd0dc3b5b5a7dc526e238a0fb8fff656925805ef
3
+ size 19017784