RichardErkhov commited on
Commit
e804090
1 Parent(s): 4bf1fce

uploaded model

Browse files
Files changed (1) hide show
  1. config.json +63 -0
config.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "hebrew-bad_wiki-gpt_neo-tiny",
3
+ "activation_function": "gelu_new",
4
+ "architectures": [
5
+ "GPTNeoForCausalLM"
6
+ ],
7
+ "attention_dropout": 0.1,
8
+ "attention_layers": [
9
+ "global",
10
+ "global",
11
+ "global",
12
+ "global",
13
+ "global",
14
+ "global"
15
+ ],
16
+ "attention_types": [
17
+ [
18
+ [
19
+ "global"
20
+ ],
21
+ 6
22
+ ]
23
+ ],
24
+ "bos_token_id": 50256,
25
+ "classifier_dropout": 0.1,
26
+ "embed_dropout": 0.1,
27
+ "eos_token_id": 50256,
28
+ "gradient_checkpointing": false,
29
+ "hidden_size": 768,
30
+ "initializer_range": 0.02,
31
+ "intermediate_size": null,
32
+ "layer_norm_epsilon": 1e-05,
33
+ "max_position_embeddings": 1024,
34
+ "model_type": "gpt_neo",
35
+ "num_heads": 12,
36
+ "num_layers": 6,
37
+ "quantization_config": {
38
+ "_load_in_4bit": false,
39
+ "_load_in_8bit": true,
40
+ "bnb_4bit_compute_dtype": "float32",
41
+ "bnb_4bit_quant_storage": "uint8",
42
+ "bnb_4bit_quant_type": "fp4",
43
+ "bnb_4bit_use_double_quant": false,
44
+ "llm_int8_enable_fp32_cpu_offload": false,
45
+ "llm_int8_has_fp16_weight": false,
46
+ "llm_int8_skip_modules": null,
47
+ "llm_int8_threshold": 6.0,
48
+ "load_in_4bit": false,
49
+ "load_in_8bit": true,
50
+ "quant_method": "bitsandbytes"
51
+ },
52
+ "resid_dropout": 0.1,
53
+ "summary_activation": null,
54
+ "summary_first_dropout": 0.1,
55
+ "summary_proj_to_labels": true,
56
+ "summary_type": "cls_index",
57
+ "summary_use_proj": true,
58
+ "torch_dtype": "float16",
59
+ "transformers_version": "4.39.3",
60
+ "use_cache": true,
61
+ "vocab_size": 50257,
62
+ "window_size": 256
63
+ }