初始化项目,由ModelHub XC社区提供模型
Model: Entrit/Qwen2.5-7B-trit-uniform-d1 Source: Original Platform
This commit is contained in:
21
quantization_config.json
Normal file
21
quantization_config.json
Normal file
@@ -0,0 +1,21 @@
|
||||
{
|
||||
"model": "qwen2.5-7B",
|
||||
"codec": "uniform-d1",
|
||||
"bpw": 1.882142969606371,
|
||||
"trit_bpw": 1.5849625007211532,
|
||||
"scale_bpw": 0.29718046888521776,
|
||||
"depth_pcts": {
|
||||
"1": 1.0,
|
||||
"2": 0.0,
|
||||
"3": 0.0,
|
||||
"4": 0.0
|
||||
},
|
||||
"n_matrices": 196,
|
||||
"group_size": 16,
|
||||
"lm_head_skipped": true,
|
||||
"codec_params": {
|
||||
"mode": "uniform",
|
||||
"scale_depth": 3,
|
||||
"depth": 1
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user