README.md
3.5 KB · 45 lines · markdown Raw
1 ---
2 base_model:
3 - tencent/WeVisCut
4 library_name: transformers
5 tags:
6 - text-generation-inference
7 - llama-cpp
8 - qwen3_5
9 - video-editing
10 - agent
11 - tool-use
12 - long-horizon
13 - reinforcement-learning
14 - qwen3.5
15 license: apache-2.0
16 language:
17 - en
18 pipeline_tag: image-text-to-text
19 ---
20
21 # **tencent-WeVisCut-GGUF**
22
23 > **WeVisCut** is a 9B tool-using video-editing agent from Tencent, built on `Qwen3.5-9B`, that converts knowledge articles and narration scripts into multi-minute edited videos. Rather than generating videos directly, it uses 12 tools for planning, narration, asset retrieval, composition, inspection, and repair, with a separate runtime executing editing workflows. Trained on executable editing trajectories, it also supports reinforcement learning. Its Studio UI provides an editable timeline, manual and chat-based revisions, per-shot rerendering, pause/resume, and version rollback. The model card provides no quantitative benchmark results; output quality depends on configured retrieval, TTS, rendering services, and templates. BF16 inference requires approximately 24 GB GPU memory and Python 3.11+. The runtime is Apache-2.0, while the model weights carry an `other` license and separate usage terms. Intended for research and production assistance, generated videos should be reviewed for factual accuracy and quality before publication.
24
25 > [!NOTE]
26 > MTP — Speculative Decoding: No
27
28 ## Model Files
29
30 | File Name | Quant Type | File Size | File Link | Description |
31 |-----------|------------|-----------|-----------|-------------|
32 | WeVisCut.BF16.gguf | BF16 | 17.9 GB | [Link](https://huggingface.co/prithivMLmods/tencent-WeVisCut-GGUF/blob/main/WeVisCut.BF16.gguf) | Full BF16 weights. Highest quality, largest file size. |
33 | WeVisCut.Q3_K_L.gguf | Q3_K_L | 4.93 GB | [Link](https://huggingface.co/prithivMLmods/tencent-WeVisCut-GGUF/blob/main/WeVisCut.Q3_K_L.gguf) | Lower quality but usable, good for low RAM availability. |
34 | WeVisCut.Q3_K_M.gguf | Q3_K_M | 4.62 GB | [Link](https://huggingface.co/prithivMLmods/tencent-WeVisCut-GGUF/blob/main/WeVisCut.Q3_K_M.gguf) | Low quality. |
35 | WeVisCut.Q4_K_M.gguf | Q4_K_M | 5.63 GB | [Link](https://huggingface.co/prithivMLmods/tencent-WeVisCut-GGUF/blob/main/WeVisCut.Q4_K_M.gguf) | Good quality, default size for most use cases, *recommended*. |
36 | WeVisCut.Q4_K_S.gguf | Q4_K_S | 5.35 GB | [Link](https://huggingface.co/prithivMLmods/tencent-WeVisCut-GGUF/blob/main/WeVisCut.Q4_K_S.gguf) | Slightly lower quality with more space savings, *recommended*. |
37 | WeVisCut.Q5_K_M.gguf | Q5_K_M | 6.47 GB | [Link](https://huggingface.co/prithivMLmods/tencent-WeVisCut-GGUF/blob/main/WeVisCut.Q5_K_M.gguf) | High quality, *recommended*. |
38 | WeVisCut.Q5_K_S.gguf | Q5_K_S | 6.31 GB | [Link](https://huggingface.co/prithivMLmods/tencent-WeVisCut-GGUF/blob/main/WeVisCut.Q5_K_S.gguf) | High quality, *recommended*. |
39 | WeVisCut.Q6_K.gguf | Q6_K | 7.36 GB | [Link](https://huggingface.co/prithivMLmods/tencent-WeVisCut-GGUF/blob/main/WeVisCut.Q6_K.gguf) | Very high quality, near perfect, *recommended*. |
40 | WeVisCut.Q8_0.gguf | Q8_0 | 9.53 GB | [Link](https://huggingface.co/prithivMLmods/tencent-WeVisCut-GGUF/blob/main/WeVisCut.Q8_0.gguf) | Extremely high quality, generally unneeded but max available quant. |
41 | WeVisCut.mmproj-bf16.gguf | mmproj-bf16 | 922 MB | [Link](https://huggingface.co/prithivMLmods/tencent-WeVisCut-GGUF/blob/main/WeVisCut.mmproj-bf16.gguf) | Multimodal projection file in BF16 format. Used for vision/language models. |
42
43 ## llama.cpp
44
45 LLM inference in C/C++ — https://github.com/ggml-org/llama.cpp