tianzl66/Llama-3.1-8B-Instruct-Magicoder-50K-LoRA-E1

Updated Aug 27

Note LoRA baseline after epoch 1.

tianzl66/Llama-3.1-8B-Instruct-Magicoder-50K-SpectralSurgery-HNS4p1

Updated Aug 27

Note Spectral Surgery with HNS 4 fast + 1 stable steps.

tianzl66/Llama-3.1-8B-Instruct-Magicoder-50K-SpectralSurgery-HNS8p2

Updated Aug 27 • 14

Note Spectral Surgery with HNS 8 fast + 2 stable steps.

tianzl66/Qwen3-8B-Magicoder-50K-LoRA-E1

Updated Aug 28 • 26

Note Qwen3-8B Magicoder-50K epoch-1 LoRA baseline: HumanEval 67.07%, MBPP 72.76%.

tianzl66/Qwen3-8B-Magicoder-50K-SpectralSurgery-HNS4p1-AllMods

Updated Aug 28 • 19

Note Epoch-1 LoRA + Spectral Surgery HNS 4+1 on all modules: HumanEval 74.39%, MBPP 75.10%.

tianzl66/Qwen3-8B-Magicoder-50K-LoRA-E1-Seed43

Text Generation • Updated 22 days ago • 39

Note B300 training seed43; source LoRA; actual warmup0. Full configuration, in-domain HNS grid, hashes and reconstruction code included.

tianzl66/Qwen3-8B-Magicoder-50K-LoRA-E1-Seed44

Text Generation • Updated 22 days ago • 18

Note B300 training seed44; source LoRA; actual warmup0. Full configuration, in-domain HNS grid, hashes and reconstruction code included.

tianzl66/Llama-3.1-8B-Instruct-Magicoder-50K-LoRA-E1-Seed43

Text Generation • Updated 22 days ago • 21

Note B300 training seed43; source LoRA; actual warmup0. Full configuration, in-domain HNS grid, hashes and reconstruction code included.

tianzl66/Llama-3.1-8B-Instruct-Magicoder-50K-LoRA-E1-Seed44

Text Generation • Updated 22 days ago • 22

Note B300 training seed44; source LoRA; actual warmup0. Full configuration, in-domain HNS grid, hashes and reconstruction code included.