upload auto_gptq format
Browse filesSigned-off-by: n1ck-guo <heng.guo@intel.com>
- config.json +5 -4
- quantize_config.json +25 -0
config.json
CHANGED
@@ -26,11 +26,11 @@
|
|
26 |
"quantization_config": {
|
27 |
"amp": true,
|
28 |
"autoround_version": "0.4.2.dev",
|
29 |
-
"backend": "auto_round:gptq:exllamav2",
|
30 |
"batch_size": 8,
|
31 |
"bits": 4,
|
|
|
32 |
"data_type": "int",
|
33 |
-
"
|
34 |
"enable_minmax_tuning": true,
|
35 |
"enable_norm_bias_tuning": false,
|
36 |
"enable_quanted_input": true,
|
@@ -41,11 +41,12 @@
|
|
41 |
"lr": 0.001,
|
42 |
"minmax_lr": 0.001,
|
43 |
"nsamples": 512,
|
44 |
-
"quant_method": "
|
45 |
"scale_dtype": "torch.float16",
|
46 |
"seqlen": 2048,
|
47 |
"sym": true,
|
48 |
-
"to_quant_block_names": null
|
|
|
49 |
},
|
50 |
"rms_norm_eps": 1e-05,
|
51 |
"rope_scaling": {
|
|
|
26 |
"quantization_config": {
|
27 |
"amp": true,
|
28 |
"autoround_version": "0.4.2.dev",
|
|
|
29 |
"batch_size": 8,
|
30 |
"bits": 4,
|
31 |
+
"damp_percent": 0.01,
|
32 |
"data_type": "int",
|
33 |
+
"desc_act": false,
|
34 |
"enable_minmax_tuning": true,
|
35 |
"enable_norm_bias_tuning": false,
|
36 |
"enable_quanted_input": true,
|
|
|
41 |
"lr": 0.001,
|
42 |
"minmax_lr": 0.001,
|
43 |
"nsamples": 512,
|
44 |
+
"quant_method": "gptq",
|
45 |
"scale_dtype": "torch.float16",
|
46 |
"seqlen": 2048,
|
47 |
"sym": true,
|
48 |
+
"to_quant_block_names": null,
|
49 |
+
"true_sequential": false
|
50 |
},
|
51 |
"rms_norm_eps": 1e-05,
|
52 |
"rope_scaling": {
|
quantize_config.json
ADDED
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bits": 4,
|
3 |
+
"group_size": 128,
|
4 |
+
"sym": true,
|
5 |
+
"data_type": "int",
|
6 |
+
"enable_quanted_input": true,
|
7 |
+
"enable_minmax_tuning": true,
|
8 |
+
"seqlen": 2048,
|
9 |
+
"batch_size": 8,
|
10 |
+
"scale_dtype": "torch.float16",
|
11 |
+
"lr": 0.001,
|
12 |
+
"minmax_lr": 0.001,
|
13 |
+
"gradient_accumulate_steps": 1,
|
14 |
+
"iters": 1000,
|
15 |
+
"amp": true,
|
16 |
+
"nsamples": 512,
|
17 |
+
"low_gpu_mem_usage": true,
|
18 |
+
"to_quant_block_names": null,
|
19 |
+
"enable_norm_bias_tuning": false,
|
20 |
+
"autoround_version": "0.4.2.dev",
|
21 |
+
"quant_method": "gptq",
|
22 |
+
"desc_act": false,
|
23 |
+
"true_sequential": false,
|
24 |
+
"damp_percent": 0.01
|
25 |
+
}
|