PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度
Commits on main
support minimax_m2 ignore layer: block_sparse_moe.gate (#1508)
a32224d Xin He committed 8 months ago
Update README.md: Add nightly installation instructions for auto-round (#1505)
1bca8f7 chen, suyue committed 8 months ago
Add handle_generation_config function to manage model generation_config saving failure (#1448)
32c1e3c Xin He committed 8 months ago
Update MOE linear_loop implementation for speedup and matching name in vLLM (#1478)
d150118 Xin He committed 8 months ago
gguf better support for transformers5.0 and fix bug of Qwen3Next (#1474)
ab14698 Heng Guo committed 8 months ago
Fix HPU CI: apply temporary version restriction on compressed_tensors (#1479)
81d27d2 Sun, Xuehao committed 8 months ago
Add envs.py for environment checks and update imports in test files (#1432)
388e6c0 Xin He committed 8 months ago
reduce AutoGPTQ priority and add transformer requirement (#1467)
e7e1c78 Xin He committed 8 months ago
fix: qwen3-next NVFP4 quantization returns act_max incorrectly (#1470)
229d1fe Xin He committed 8 months ago
Fix KeyError when GPU is missing from accelerate max_memory (#1457)
b697e1d Liang Lv committed 8 months ago
support multiple device evaluation for activation quantized model (#1394)
bd4241d Wenhua Cheng committed 8 months ago
fix: preserve classmethod descriptor in from_pretrained monkey patch (#1460)
0d0f836 Yi Liu committed 8 months ago
update logic for compatibility with gptqmodel:exllamav2 backend (#1438)
86231c4 Xin He committed 9 months ago