Isaachh commited on
Commit
7adf7c2
1 Parent(s): 844e12e
Files changed (4) hide show
  1. README.md +17 -0
  2. config.json +46 -0
  3. mm_projector.bin +3 -0
  4. trainer_state.json +0 -0
README.md ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ inference: false
3
+ license: apache-2.0
4
+ ---
5
+
6
+ # Model Card
7
+
8
+ Bunny is a family of lightweight multimodal models.
9
+
10
+ Bunny-pretrain-phi-3-siglip is the pretrained weights for [Bunny-v1.0-4B](https://huggingface.co/BAAI/Bunny-v1_0-4B), which leverages Phi-3-Mini-4K-Instruct as the language model backbone and SigLIP as the vision encoder.
11
+ It is pretrained on LAION-2M.
12
+
13
+ More details about this model can be found in [GitHub](https://github.com/BAAI-DCAI/Bunny).
14
+
15
+ # License
16
+ This project utilizes certain datasets and checkpoints that are subject to their respective original licenses. Users must comply with all terms and conditions of these original licenses.
17
+ The content of this project itself is licensed under the Apache license 2.0.
config.json ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "bunny-pretrain-phi-3-siglip",
3
+ "architectures": [
4
+ "Phi3ForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "auto_map": {
8
+ "AutoConfig": "configuration_phi3.Phi3Config",
9
+ "AutoModelForCausalLM": "modeling_phi3.Phi3ForCausalLM"
10
+ },
11
+ "bos_token_id": 1,
12
+ "embd_pdrop": 0.0,
13
+ "eos_token_id": 32000,
14
+ "freeze_mm_mlp_adapter": false,
15
+ "hidden_act": "silu",
16
+ "hidden_size": 3072,
17
+ "image_aspect_ratio": "square",
18
+ "initializer_range": 0.02,
19
+ "intermediate_size": 8192,
20
+ "max_position_embeddings": 4096,
21
+ "mm_hidden_size": 1152,
22
+ "mm_projector_lr": null,
23
+ "mm_projector_type": "mlp2x_gelu",
24
+ "mm_vision_tower": "google/siglip-so400m-patch14-384",
25
+ "model_type": "bunny-phi3",
26
+ "num_attention_heads": 32,
27
+ "num_hidden_layers": 32,
28
+ "num_key_value_heads": 32,
29
+ "original_max_position_embeddings": 4096,
30
+ "pad_token_id": 32000,
31
+ "resid_pdrop": 0.0,
32
+ "rms_norm_eps": 1e-05,
33
+ "rope_scaling": null,
34
+ "rope_theta": 10000.0,
35
+ "sliding_window": 2047,
36
+ "tie_word_embeddings": false,
37
+ "tokenizer_model_max_length": 2048,
38
+ "tokenizer_padding_side": "right",
39
+ "torch_dtype": "bfloat16",
40
+ "transformers_version": "4.40.0",
41
+ "tune_mm_mlp_adapter": true,
42
+ "unfreeze_vision_tower": false,
43
+ "use_cache": true,
44
+ "use_mm_proj": true,
45
+ "vocab_size": 32064
46
+ }
mm_projector.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f86fb430bfa74ba28841af0aecbf339fe9bcff7933b4d636641bc61c0c7c4776
3
+ size 25966648
trainer_state.json ADDED
The diff for this file is too large to render. See raw diff