FradSer commited on
Commit
f8e03d7
·
1 Parent(s): 0b0e570

Revert "Trained with Unsloth"

Browse files

This reverts commit 31e9b0186fb64ae921d7590ee0ad51dbe45d3c07.

Files changed (4) hide show
  1. README.md +0 -1
  2. config.json +0 -33
  3. generation_config.json +0 -8
  4. pytorch_model.bin +0 -3
README.md CHANGED
@@ -7,7 +7,6 @@ tags:
7
  - unsloth
8
  - qwen2
9
  - trl
10
- - sft
11
  license: apache-2.0
12
  language:
13
  - en
 
7
  - unsloth
8
  - qwen2
9
  - trl
 
10
  license: apache-2.0
11
  language:
12
  - en
config.json DELETED
@@ -1,33 +0,0 @@
1
- {
2
- "_name_or_path": "unsloth/deepseek-r1-distill-qwen-1.5b-bnb-4bit",
3
- "architectures": [
4
- "Qwen2ForCausalLM"
5
- ],
6
- "attention_dropout": 0.0,
7
- "bos_token_id": 151646,
8
- "eos_token_id": 151643,
9
- "hidden_act": "silu",
10
- "hidden_size": 1536,
11
- "initializer_range": 0.02,
12
- "intermediate_size": 8960,
13
- "max_position_embeddings": 131072,
14
- "max_window_layers": 21,
15
- "model_type": "qwen2",
16
- "num_attention_heads": 12,
17
- "num_hidden_layers": 28,
18
- "num_key_value_heads": 2,
19
- "pad_token_id": 151654,
20
- "rms_norm_eps": 1e-06,
21
- "rope_scaling": null,
22
- "rope_theta": 10000,
23
- "sliding_window": null,
24
- "tie_word_embeddings": false,
25
- "torch_dtype": "float16",
26
- "transformers_version": "4.48.3",
27
- "unsloth_fixed": true,
28
- "unsloth_version": "2025.3.10",
29
- "use_cache": true,
30
- "use_mrope": false,
31
- "use_sliding_window": false,
32
- "vocab_size": 151936
33
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
generation_config.json DELETED
@@ -1,8 +0,0 @@
1
- {
2
- "_from_model_config": true,
3
- "bos_token_id": 151646,
4
- "eos_token_id": 151643,
5
- "max_length": 131072,
6
- "pad_token_id": 0,
7
- "transformers_version": "4.48.3"
8
- }
 
 
 
 
 
 
 
 
 
pytorch_model.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:f6201315e42df6e5e1ae16e4a68f34830c92656ab859662f30513c965e7502a9
3
- size 3554278006