Updated compression_config to quantization_config
Browse files- config.json +43 -43
config.json
CHANGED
@@ -12,48 +12,6 @@
|
|
12 |
},
|
13 |
"aux_loss_alpha": 0.001,
|
14 |
"bos_token_id": 100000,
|
15 |
-
"compression_config": {
|
16 |
-
"config_groups": {
|
17 |
-
"group_0": {
|
18 |
-
"input_activations": {
|
19 |
-
"actorder": null,
|
20 |
-
"block_structure": null,
|
21 |
-
"dynamic": true,
|
22 |
-
"group_size": null,
|
23 |
-
"num_bits": 8,
|
24 |
-
"observer": "memoryless",
|
25 |
-
"observer_kwargs": {},
|
26 |
-
"strategy": "token",
|
27 |
-
"symmetric": true,
|
28 |
-
"type": "int"
|
29 |
-
},
|
30 |
-
"output_activations": null,
|
31 |
-
"targets": [
|
32 |
-
"Linear"
|
33 |
-
],
|
34 |
-
"weights": {
|
35 |
-
"actorder": null,
|
36 |
-
"block_structure": null,
|
37 |
-
"dynamic": false,
|
38 |
-
"group_size": null,
|
39 |
-
"num_bits": 8,
|
40 |
-
"observer": "minmax",
|
41 |
-
"observer_kwargs": {},
|
42 |
-
"strategy": "channel",
|
43 |
-
"symmetric": true,
|
44 |
-
"type": "int"
|
45 |
-
}
|
46 |
-
}
|
47 |
-
},
|
48 |
-
"format": "int-quantized",
|
49 |
-
"global_compression_ratio": 1.2697642717533653,
|
50 |
-
"ignore": [
|
51 |
-
"lm_head"
|
52 |
-
],
|
53 |
-
"kv_cache_scheme": null,
|
54 |
-
"quant_method": "compressed-tensors",
|
55 |
-
"quantization_status": "compressed"
|
56 |
-
},
|
57 |
"eos_token_id": 100001,
|
58 |
"ep_size": 1,
|
59 |
"first_k_dense_replace": 1,
|
@@ -99,5 +57,47 @@
|
|
99 |
"transformers_version": "4.44.2",
|
100 |
"use_cache": true,
|
101 |
"v_head_dim": 128,
|
102 |
-
"vocab_size": 102400
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
103 |
}
|
|
|
12 |
},
|
13 |
"aux_loss_alpha": 0.001,
|
14 |
"bos_token_id": 100000,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
15 |
"eos_token_id": 100001,
|
16 |
"ep_size": 1,
|
17 |
"first_k_dense_replace": 1,
|
|
|
57 |
"transformers_version": "4.44.2",
|
58 |
"use_cache": true,
|
59 |
"v_head_dim": 128,
|
60 |
+
"vocab_size": 102400,
|
61 |
+
"quantization_config": {
|
62 |
+
"config_groups": {
|
63 |
+
"group_0": {
|
64 |
+
"input_activations": {
|
65 |
+
"actorder": null,
|
66 |
+
"block_structure": null,
|
67 |
+
"dynamic": true,
|
68 |
+
"group_size": null,
|
69 |
+
"num_bits": 8,
|
70 |
+
"observer": "memoryless",
|
71 |
+
"observer_kwargs": {},
|
72 |
+
"strategy": "token",
|
73 |
+
"symmetric": true,
|
74 |
+
"type": "int"
|
75 |
+
},
|
76 |
+
"output_activations": null,
|
77 |
+
"targets": [
|
78 |
+
"Linear"
|
79 |
+
],
|
80 |
+
"weights": {
|
81 |
+
"actorder": null,
|
82 |
+
"block_structure": null,
|
83 |
+
"dynamic": false,
|
84 |
+
"group_size": null,
|
85 |
+
"num_bits": 8,
|
86 |
+
"observer": "minmax",
|
87 |
+
"observer_kwargs": {},
|
88 |
+
"strategy": "channel",
|
89 |
+
"symmetric": true,
|
90 |
+
"type": "int"
|
91 |
+
}
|
92 |
+
}
|
93 |
+
},
|
94 |
+
"format": "int-quantized",
|
95 |
+
"global_compression_ratio": 1.2697642717533653,
|
96 |
+
"ignore": [
|
97 |
+
"lm_head"
|
98 |
+
],
|
99 |
+
"kv_cache_scheme": null,
|
100 |
+
"quant_method": "compressed-tensors",
|
101 |
+
"quantization_status": "compressed"
|
102 |
+
}
|
103 |
}
|