LoRA adapters trained for 5 progressively shorter chain-of-thought styles on GSM8K, plus the eval artifacts behind the Pareto curve.
Frolov Anatolii
ssurface
·
AI & ML interests
None yet
Recent Activity
updated a collection 7 days ago
Qwen3-4B CoT Compression Study updated a collection 7 days ago
Qwen3-4B CoT Compression Study updated a collection 7 days ago
Qwen3-4B CoT Compression StudyOrganizations
models 9
ssurface/qwen3-4b-cot-compress-l5
Text Generation • Updated • 28
ssurface/qwen3-4b-cot-compress-l4
Text Generation • Updated • 27
ssurface/qwen3-4b-cot-compress-l3
Text Generation • Updated • 27
ssurface/qwen3-4b-cot-compress-l2
Text Generation • Updated • 22
ssurface/qwen3-4b-cot-compress-l1
Text Generation • Updated • 26
ssurface/paradetox-gemma-4b-bnb
Updated
ssurface/qwen2.5-1.5B-DimABSA2026-LORA
Text Generation • Updated
ssurface/mmBERT-base-finetuned-dimabsa
Updated
ssurface/dimabsa
0.4B • Updated • 2