初始化项目,由ModelHub XC社区提供模型
Model: AvoCahDoe/llama-2-7b-rlmpq-balanced Source: Original Platform
This commit is contained in:
23
rlmpq_metrics.json
Normal file
23
rlmpq_metrics.json
Normal file
@@ -0,0 +1,23 @@
|
||||
{
|
||||
"framework": "RL-NMP-Model-Quantasation",
|
||||
"method": "RL-MPQ",
|
||||
"repo_id": "AvoCahDoe/llama-2-7b-rlmpq-balanced",
|
||||
"base_model": "meta-llama/Llama-2-7b-hf",
|
||||
"model_slug": "llama_2_7b",
|
||||
"scenario": "Balanced",
|
||||
"scenario_label": "Balanced",
|
||||
"avg_bits_per_weight": 4.375,
|
||||
"compression_vs_fp16": 3.6571,
|
||||
"wikitext2_perplexity": 5.0437,
|
||||
"bit_distribution": {
|
||||
"4": 29,
|
||||
"8": 3
|
||||
},
|
||||
"quantization": {
|
||||
"type": "fake-quant-fp16",
|
||||
"group_size": 128,
|
||||
"scheme": "per-layer asymmetric group-wise",
|
||||
"packed_format": false
|
||||
},
|
||||
"exported_at": "2026-06-11T16:40:33.183918"
|
||||
}
|
||||
Reference in New Issue
Block a user