OPEA
/

Safetensors
qwen2_vl
4-bit precision
intel/auto-round
File size: 588 Bytes
e67cae7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
61faab7
e67cae7
61faab7
 
e67cae7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
{
  "bits": 4,
  "group_size": 128,
  "sym": true,
  "data_type": "int",
  "enable_quanted_input": true,
  "enable_minmax_tuning": true,
  "seqlen": 2048,
  "batch_size": 8,
  "scale_dtype": "torch.float16",
  "lr": 0.001,
  "minmax_lr": 0.001,
  "gradient_accumulate_steps": 1,
  "iters": 1000,
  "amp": true,
  "nsamples": 512,
  "low_gpu_mem_usage": true,
  "to_quant_block_names": "model.layers",
  "enable_norm_bias_tuning": false,
  "dataset": "NeelNanda/pile-10k",
  "autoround_version": "0.4.2.dev",
  "quant_method": "intel/auto-round",
  "backend": "auto_round:gptq:exllamav2"
}