danielhanchen commited on
Commit
92a21d6
1 Parent(s): 4f7464e

Upload config

Browse files
Files changed (2) hide show
  1. README.md +1 -2
  2. config.json +5 -2
README.md CHANGED
@@ -1,14 +1,13 @@
1
  ---
2
  language:
3
  - en
4
- license: apache-2.0
5
  library_name: transformers
 
6
  tags:
7
  - unsloth
8
  - transformers
9
  - zephyr
10
  - sft
11
-
12
  ---
13
 
14
  # Finetune Mistral, Gemma, Llama 2-5x faster with 70% less memory via Unsloth!
 
1
  ---
2
  language:
3
  - en
 
4
  library_name: transformers
5
+ license: apache-2.0
6
  tags:
7
  - unsloth
8
  - transformers
9
  - zephyr
10
  - sft
 
11
  ---
12
 
13
  # Finetune Mistral, Gemma, Llama 2-5x faster with 70% less memory via Unsloth!
config.json CHANGED
@@ -1,11 +1,12 @@
1
  {
2
- "_name_or_path": "HuggingFaceH4/mistral-7b-sft-beta",
3
  "architectures": [
4
  "MistralForCausalLM"
5
  ],
6
  "attention_dropout": 0.0,
7
  "bos_token_id": 1,
8
  "eos_token_id": 2,
 
9
  "hidden_act": "silu",
10
  "hidden_size": 4096,
11
  "initializer_range": 0.02,
@@ -15,12 +16,14 @@
15
  "num_attention_heads": 32,
16
  "num_hidden_layers": 32,
17
  "num_key_value_heads": 8,
 
18
  "rms_norm_eps": 1e-05,
19
  "rope_theta": 10000.0,
20
  "sliding_window": 4096,
21
  "tie_word_embeddings": false,
22
  "torch_dtype": "bfloat16",
23
- "transformers_version": "4.37.0.dev0",
 
24
  "use_cache": true,
25
  "vocab_size": 32000
26
  }
 
1
  {
2
+ "_name_or_path": "unsloth/zephyr-sft",
3
  "architectures": [
4
  "MistralForCausalLM"
5
  ],
6
  "attention_dropout": 0.0,
7
  "bos_token_id": 1,
8
  "eos_token_id": 2,
9
+ "head_dim": 128,
10
  "hidden_act": "silu",
11
  "hidden_size": 4096,
12
  "initializer_range": 0.02,
 
16
  "num_attention_heads": 32,
17
  "num_hidden_layers": 32,
18
  "num_key_value_heads": 8,
19
+ "pad_token_id": 0,
20
  "rms_norm_eps": 1e-05,
21
  "rope_theta": 10000.0,
22
  "sliding_window": 4096,
23
  "tie_word_embeddings": false,
24
  "torch_dtype": "bfloat16",
25
+ "transformers_version": "4.44.0",
26
+ "unsloth_version": "2024.8",
27
  "use_cache": true,
28
  "vocab_size": 32000
29
  }