TimeLens-8B-GGUF
TimeLens-8B from TencentARC is an 8B-parameter multimodal vision-language model fine-tuned from Qwen3-VL-8B-Instruct using a novel RLVR (reinforcement learning with verifiable rewards) recipe on the high-quality TimeLens-100K VTG dataset, achieving state-of-the-art video temporal grounding performance among open-source models with 72.0% [email protected] (Charades-TimeLens), 64.5% [email protected] (ActivityNet-TimeLens), and 75.6% [email protected] (QVHighlights-TimeLens), significantly outperforming baselines like Qwen3-VL-8B-Instruct and Qwen2.5-VL-7B. Designed for precise localization of visual events described by natural language queries, it outputs timestamped segments in the format "The event happens in - seconds" using low FPS=2 sampling (min_pixels=642828, total_pixels=143362828) for efficient video processing via Transformers with Flash-Attention-2 support. Released with code, project page, and TimeLens-Bench evaluation suite, it excels on Charades-TimeLens, ActivityNet-TimeLens, and QVHighlights-TimeLens leaderboards for research in video understanding, temporal reasoning, and event detection.
TimeLens-8B [GGUF]
| File Name | Quant Type | File Size | File Link |
|---|---|---|---|
| TimeLens-8B.IQ4_XS.gguf | IQ4_XS | 4.59 GB | Download |
| TimeLens-8B.Q2_K.gguf | Q2_K | 3.28 GB | Download |
| TimeLens-8B.Q3_K_L.gguf | Q3_K_L | 4.43 GB | Download |
| TimeLens-8B.Q3_K_M.gguf | Q3_K_M | 4.12 GB | Download |
| TimeLens-8B.Q3_K_S.gguf | Q3_K_S | 3.77 GB | Download |
| TimeLens-8B.Q4_K_M.gguf | Q4_K_M | 5.03 GB | Download |
| TimeLens-8B.Q4_K_S.gguf | Q4_K_S | 4.8 GB | Download |
| TimeLens-8B.Q5_K_M.gguf | Q5_K_M | 5.85 GB | Download |
| TimeLens-8B.Q5_K_S.gguf | Q5_K_S | 5.72 GB | Download |
| TimeLens-8B.Q6_K.gguf | Q6_K | 6.73 GB | Download |
| TimeLens-8B.Q8_0.gguf | Q8_0 | 8.71 GB | Download |
| TimeLens-8B.f16.gguf | F16 | 16.4 GB | Download |
| TimeLens-8B.mmproj-Q8_0.gguf | mmproj-Q8_0 | 752 MB | Download |
| TimeLens-8B.mmproj-f16.gguf | mmproj-f16 | 1.16 GB | Download |
Quants Usage
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
Here is a handy graph by ikawrakow comparing some lower-quality quant types (lower is better):
- Downloads last month
- 277
2-bit
3-bit
4-bit
5-bit
6-bit
8-bit
16-bit