akshitha-k commited on
Commit
7a1d694
1 Parent(s): f896a1b

Update config.json

Browse files
Files changed (1) hide show
  1. config.json +13 -12
config.json CHANGED
@@ -1,14 +1,6 @@
1
  {
2
- "_name_or_path": "unsloth/gemma-2-9b",
3
- "architectures": [
4
- "Gemma2ForCausalLM"
5
- ],
6
- "quantization": [
7
- "bits": 4,
8
- "dtype": "float16",
9
- "is_dynamic": true,
10
- "low_cpu_mem_usage": true
11
- ]
12
  "attention_bias": false,
13
  "attention_dropout": 0.0,
14
  "attn_logit_softcapping": 50.0,
@@ -18,10 +10,11 @@
18
  "final_logit_softcapping": 30.0,
19
  "head_dim": 256,
20
  "hidden_act": "gelu_pytorch_tanh",
21
- "hidden_activation": "gelu_pytorch_tanh",
22
  "hidden_size": 3584,
23
  "initializer_range": 0.02,
24
  "intermediate_size": 14336,
 
 
25
  "max_position_embeddings": 8192,
26
  "model_type": "gemma2",
27
  "num_attention_heads": 16,
@@ -33,9 +26,17 @@
33
  "rope_theta": 10000.0,
34
  "sliding_window": 4096,
35
  "sliding_window_size": 4096,
 
36
  "torch_dtype": "bfloat16",
37
  "transformers_version": "4.44.2",
38
  "unsloth_version": "2024.9",
39
  "use_cache": true,
40
- "vocab_size": 256000
 
 
 
 
 
 
 
41
  }
 
1
  {
2
+ "_name_or_path": "akshitha-k/oneliner-to-stories",
3
+ "architectures": ["Gemma2ForCausalLM"],
 
 
 
 
 
 
 
 
4
  "attention_bias": false,
5
  "attention_dropout": 0.0,
6
  "attn_logit_softcapping": 50.0,
 
10
  "final_logit_softcapping": 30.0,
11
  "head_dim": 256,
12
  "hidden_act": "gelu_pytorch_tanh",
 
13
  "hidden_size": 3584,
14
  "initializer_range": 0.02,
15
  "intermediate_size": 14336,
16
+ "lora_alpha": 40,
17
+ "lora_dropout": 0,
18
  "max_position_embeddings": 8192,
19
  "model_type": "gemma2",
20
  "num_attention_heads": 16,
 
26
  "rope_theta": 10000.0,
27
  "sliding_window": 4096,
28
  "sliding_window_size": 4096,
29
+ "target_modules": ["q_proj", "k_proj", "v_proj", "o_proj", "gate_proj", "up_proj", "down_proj"],
30
  "torch_dtype": "bfloat16",
31
  "transformers_version": "4.44.2",
32
  "unsloth_version": "2024.9",
33
  "use_cache": true,
34
+ "vocab_size": 256000,
35
+ "quantization": {
36
+ "approach": "dynamic",
37
+ "bits": 4,
38
+ "dtype": "float16",
39
+ "is_dynamic": true,
40
+ "low_cpu_mem_usage": true
41
+ }
42
  }