-
TinyV: Reducing False Negatives in Verification Improves RL for LLM Reasoning
Paper • 2505.14625 • Published • 13 -
1
TinyV
💬Verify model answers against ground truth
-
zhangchenxu/TinyV-Qwen3-1.7B
Text Generation • 2B • Updated • 7 -
zhangchenxu/TinyV-Qwen3-1.7B-Think
Text Generation • 2B • Updated • 8 • 1
Zhangchen Xu PRO
zhangchenxu
AI & ML interests
LLM Data, Alignment, Post-Training, Safety
Recent Activity
published
a model
about 11 hours ago
zhangchenxu/RB-Qwen2.5-VL-7B-Instruct-vlr_syn_filtered_10k_exp6_hybrid_nothink_20k-GRPO_step80
updated
a model
about 11 hours ago
zhangchenxu/RB-Qwen2.5-VL-7B-Instruct-vlr_syn_filtered_10k_exp6_hybrid_nothink_20k-GRPO_step80
updated
a model
about 13 hours ago
zhangchenxu/RB-Qwen2.5-VL-7B-Instruct-vlr_syn_filtered_10k_exp6_hybrid_nothink-GRPO_step160
Organizations
-
TinyV: Reducing False Negatives in Verification Improves RL for LLM Reasoning
Paper • 2505.14625 • Published • 13 -
1
TinyV
💬Verify model answers against ground truth
-
zhangchenxu/TinyV-Qwen3-1.7B
Text Generation • 2B • Updated • 7 -
zhangchenxu/TinyV-Qwen3-1.7B-Think
Text Generation • 2B • Updated • 8 • 1
models
17
zhangchenxu/RB-Qwen2.5-VL-7B-Instruct-vlr_syn_filtered_10k_exp6_hybrid_nothink_20k-GRPO_step80
8B
•
Updated
•
1
zhangchenxu/RB-Qwen2.5-VL-7B-Instruct-vlr_syn_filtered_10k_exp6_hybrid_nothink-GRPO_step160
8B
•
Updated
•
1
zhangchenxu/RB-Qwen2.5-VL-7B-Instruct-vlr_syn_filtered_10k_exp6_hybrid_nothink-GRPO_step240
8B
•
Updated
•
1
zhangchenxu/RB-Qwen2.5-VL-7B-Instruct-vlr_syn_filtered_10k_exp6_hybrid_nothink-GRPO_step304
8B
•
Updated
•
1
zhangchenxu/RB-Qwen2.5-VL-7B-Instruct-vlr_syn_filtered_10k_exp6_hybrid_nothink-GRPO_step256
8B
•
Updated
•
1
zhangchenxu/RB-Qwen2.5-VL-7B-Instruct-vlr_syn_filtered_10k_exp6_hybrid_nothink-GRPO_step144
8B
•
Updated
•
74
zhangchenxu/Qwen2.5-VL-7B-Instruct-SFT-visualsphinx_10k_random-LR2.0e-5-EPOCHS3-LF
Image-to-Text
•
8B
•
Updated
•
23
zhangchenxu/Qwen2.5-VL-7B-Instruct-SFT-visualsphinx_10k_reject-LR2.0e-5-EPOCHS3-LF
Image-to-Text
•
8B
•
Updated
•
24
zhangchenxu/Qwen2.5-VL-7B-Instruct-SFT-visualsphinx_10k_random-LR2.0e-5-EPOCHS2-LF-10000000
Updated
zhangchenxu/TinyV-1.5B-Think
Text Generation
•
2B
•
Updated
•
20
datasets
14
zhangchenxu/HardVerify-Math
Viewer
•
Updated
•
250
•
55
•
1
zhangchenxu/TinyV_Think_Training_Data_Qwen3_Balanced
Viewer
•
Updated
•
113k
•
9
zhangchenxu/TinyV_Training_Data_Qwen3_Balanced
Viewer
•
Updated
•
113k
•
11
zhangchenxu/bigmath_tinyv_filtered
Viewer
•
Updated
•
7.01k
•
34
zhangchenxu/TinyV_Training_Data_Balanced
Viewer
•
Updated
•
159k
•
5
zhangchenxu/TinyV_Think_Training_Data_Balanced
Viewer
•
Updated
•
159k
•
4
zhangchenxu/KodCode_50K_R1
Viewer
•
Updated
•
50k
•
25
zhangchenxu/KodCode_Hard_18K_R1
Viewer
•
Updated
•
18.7k
•
12
zhangchenxu/Magpie-100k-Gemma2-9B
Viewer
•
Updated
•
100k
•
10
zhangchenxu/zero-eval
Viewer
•
Updated
•
800
•
192