RichardErkhov commited on
Commit
df99f85
·
verified ·
1 Parent(s): 0c4f029

uploaded model

Browse files
Files changed (1) hide show
  1. config.json +44 -0
config.json ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "bigscience-small-testing",
3
+ "apply_residual_connection_post_layernorm": false,
4
+ "architectures": [
5
+ "BloomForCausalLM"
6
+ ],
7
+ "attention_dropout": 0.1,
8
+ "bias_dropout_fusion": true,
9
+ "bos_token_id": 0,
10
+ "dtype": "bfloat16",
11
+ "eos_token_id": 0,
12
+ "hidden_dropout": 0.1,
13
+ "hidden_size": 64,
14
+ "initializer_range": 0.02,
15
+ "layer_norm_epsilon": 1e-05,
16
+ "masked_softmax_fusion": true,
17
+ "model_type": "bloom",
18
+ "n_head": 8,
19
+ "n_inner": null,
20
+ "n_layer": 2,
21
+ "pretraining_tp": 2,
22
+ "quantization_config": {
23
+ "_load_in_4bit": true,
24
+ "_load_in_8bit": false,
25
+ "bnb_4bit_compute_dtype": "float32",
26
+ "bnb_4bit_quant_storage": "uint8",
27
+ "bnb_4bit_quant_type": "fp4",
28
+ "bnb_4bit_use_double_quant": false,
29
+ "llm_int8_enable_fp32_cpu_offload": false,
30
+ "llm_int8_has_fp16_weight": false,
31
+ "llm_int8_skip_modules": null,
32
+ "llm_int8_threshold": 6.0,
33
+ "load_in_4bit": true,
34
+ "load_in_8bit": false,
35
+ "quant_method": "bitsandbytes"
36
+ },
37
+ "seq_length": 20,
38
+ "skip_bias_add": true,
39
+ "slow_but_exact": true,
40
+ "torch_dtype": "float16",
41
+ "transformers_version": "4.39.3",
42
+ "use_cache": false,
43
+ "vocab_size": 250880
44
+ }