Qwen3-4B checkpoints whose chat template always preserves <think> reasoning across multi-turn history, with no nonthinking mode.
Ethan Ewer
eewer
AI & ML interests
None yet
Recent Activity
updated a model 4 days ago
eewer/Qwen3-4B-Thinking-Preservation-terminus2-sft published a model 4 days ago
eewer/Qwen3-4B-Thinking-Preservation-terminus2-sft updated a dataset 7 days ago
eewer/skill-based-medium-terminus2-sftOrganizations
None yet
models 50
eewer/Qwen3-4B-Thinking-Preservation-terminus2-sft
Text Generation • 4B • Updated • 19
eewer/OpenThinkerAgent-32B-AWQ-Int4-Terminus2
Text Generation • 33B • Updated • 24
eewer/Qwen3-4B-Base-Thinking-Preservation
Text Generation • 4B • Updated • 27
eewer/Qwen3-4B-Thinking-2507-Thinking-Preservation
Text Generation • 4B • Updated • 23
eewer/Qwen3-4B-Thinking-Preservation
Text Generation • 4B • Updated • 29
eewer/Qwen3.5-4B-Thinking-Preservation
Image-Text-to-Text • 5B • Updated • 20
eewer/qwen3-vl-8b-thinking-text
8B • Updated • 12
eewer/qwen3-vl-8b-instruct-text
8B • Updated • 9
eewer/qwen3-vl-2b-thinking-text
2B • Updated • 17
eewer/qwen3-vl-2b-instruct-text
2B • Updated • 10