kuleshov commited on
Commit
22e0024
1 Parent(s): 5f201c4

first commit

Browse files
Files changed (3) hide show
  1. config.json +24 -0
  2. quant_config.json +11 -0
  3. quantized_weights.pt +3 -0
config.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "huggyllama/llama-30b",
3
+ "architectures": [
4
+ "LlamaForCausalLM"
5
+ ],
6
+ "bos_token_id": 1,
7
+ "eos_token_id": 2,
8
+ "hidden_act": "silu",
9
+ "hidden_size": 6656,
10
+ "initializer_range": 0.02,
11
+ "intermediate_size": 17920,
12
+ "max_position_embeddings": 2048,
13
+ "max_sequence_length": 2048,
14
+ "model_type": "llama",
15
+ "num_attention_heads": 52,
16
+ "num_hidden_layers": 60,
17
+ "pad_token_id": 0,
18
+ "rms_norm_eps": 1e-06,
19
+ "tie_word_embeddings": false,
20
+ "torch_dtype": "float16",
21
+ "transformers_version": "4.28.0",
22
+ "use_cache": true,
23
+ "vocab_size": 32000
24
+ }
quant_config.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": "c4",
3
+ "bits": 4,
4
+ "nsamples": 128,
5
+ "groupsize": 64,
6
+ "act_order": true,
7
+ "percdamp": 0.01,
8
+ "seed": 0,
9
+ "nearest": false,
10
+ "save": "./llama-30b-4bit"
11
+ }
quantized_weights.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0894115989321ceafdebc2cda2f8cbac48a16ae9041b01609362d9ee7e9ed3c5
3
+ size 18173219120