noahtye commited on
Commit
774f4da
·
verified ·
1 Parent(s): 9fc0328

Upload GemmaForCausalLM

Browse files
Files changed (1) hide show
  1. config.json +16 -1
config.json CHANGED
@@ -19,10 +19,25 @@
19
  "num_hidden_layers": 18,
20
  "num_key_value_heads": 1,
21
  "pad_token_id": 0,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
22
  "rms_norm_eps": 1e-06,
23
  "rope_scaling": null,
24
  "rope_theta": 10000.0,
25
- "torch_dtype": "float32",
26
  "transformers_version": "4.42.3",
27
  "use_cache": true,
28
  "vocab_size": 256000
 
19
  "num_hidden_layers": 18,
20
  "num_key_value_heads": 1,
21
  "pad_token_id": 0,
22
+ "quantization_config": {
23
+ "_load_in_4bit": true,
24
+ "_load_in_8bit": false,
25
+ "bnb_4bit_compute_dtype": "bfloat16",
26
+ "bnb_4bit_quant_storage": "uint8",
27
+ "bnb_4bit_quant_type": "nf4",
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
  "rms_norm_eps": 1e-06,
38
  "rope_scaling": null,
39
  "rope_theta": 10000.0,
40
+ "torch_dtype": "float16",
41
  "transformers_version": "4.42.3",
42
  "use_cache": true,
43
  "vocab_size": 256000