danielhanchen commited on
Commit
d74416d
·
verified ·
1 Parent(s): 9313e8f

Upload folder using huggingface_hub

Browse files
Files changed (3) hide show
  1. README.md +0 -4
  2. config.json +4 -0
  3. tokenizer_config.json +0 -0
README.md CHANGED
@@ -23,10 +23,6 @@ tags:
23
  - mistral-common
24
  - unsloth
25
  ---
26
- > [!NOTE]
27
- > Includes Unsloth **chat template fixes**! <br> For `llama.cpp`, use `--jinja`
28
- >
29
-
30
  <div>
31
  <p style="margin-top: 0;margin-bottom: 0;">
32
  <em><a href="https://docs.unsloth.ai/basics/unsloth-dynamic-v2.0-gguf">Unsloth Dynamic 2.0</a> achieves superior accuracy & outperforms other leading quants.</em>
 
23
  - mistral-common
24
  - unsloth
25
  ---
 
 
 
 
26
  <div>
27
  <p style="margin-top: 0;margin-bottom: 0;">
28
  <em><a href="https://docs.unsloth.ai/basics/unsloth-dynamic-v2.0-gguf">Unsloth Dynamic 2.0</a> achieves superior accuracy & outperforms other leading quants.</em>
config.json CHANGED
@@ -2,7 +2,9 @@
2
  "architectures": [
3
  "Mistral3ForConditionalGeneration"
4
  ],
 
5
  "torch_dtype": "bfloat16",
 
6
  "image_token_index": 10,
7
  "model_type": "mistral3",
8
  "multimodal_projector_bias": false,
@@ -11,6 +13,7 @@
11
  "spatial_merge_size": 2,
12
  "text_config": {
13
  "attention_dropout": 0.0,
 
14
  "head_dim": 128,
15
  "hidden_act": "silu",
16
  "hidden_size": 5120,
@@ -42,6 +45,7 @@
42
  "unsloth_fixed": true,
43
  "vision_config": {
44
  "attention_dropout": 0.0,
 
45
  "head_dim": 64,
46
  "hidden_act": "silu",
47
  "hidden_size": 1024,
 
2
  "architectures": [
3
  "Mistral3ForConditionalGeneration"
4
  ],
5
+ "bos_token_id": 1,
6
  "torch_dtype": "bfloat16",
7
+ "eos_token_id": 2,
8
  "image_token_index": 10,
9
  "model_type": "mistral3",
10
  "multimodal_projector_bias": false,
 
13
  "spatial_merge_size": 2,
14
  "text_config": {
15
  "attention_dropout": 0.0,
16
+ "torch_dtype": "bfloat16",
17
  "head_dim": 128,
18
  "hidden_act": "silu",
19
  "hidden_size": 5120,
 
45
  "unsloth_fixed": true,
46
  "vision_config": {
47
  "attention_dropout": 0.0,
48
+ "torch_dtype": "bfloat16",
49
  "head_dim": 64,
50
  "hidden_act": "silu",
51
  "hidden_size": 1024,
tokenizer_config.json CHANGED
The diff for this file is too large to render. See raw diff