# Video-based Text Generation on VCGBench 1.0 (test)

Source: [Video-based Text Generation on VCGBench 1.0 (test)](<https://www.sota2.com/research/sota/video-based-text-generation-on-vcgbench-1-0-test>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:38:21.389Z

Evaluation of Video-based Text Generation models using the test split of VCGBench version 1.0, assessed across five dimensions: Correct Information (CI), Detail Orientation (DO), Contextual Understanding (CU), Temporal Understanding (TU), and Consistency (CO).

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Video-based Text Generation |
| Dataset (including any stated split) | VCGBench 1.0 (test) |
| Metric | CI |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [CI](<https://www.sota2.com/research/sota/video-based-text-generation-on-vcgbench-1-0-test.md?metric=CI>): Higher is better; unit: SCALAR.

- [DO](<https://www.sota2.com/research/sota/video-based-text-generation-on-vcgbench-1-0-test.md?metric=DO>): Higher is better; unit: SCALAR.

- [CU](<https://www.sota2.com/research/sota/video-based-text-generation-on-vcgbench-1-0-test.md?metric=CU>): Higher is better; unit: SCALAR.

- [TU](<https://www.sota2.com/research/sota/video-based-text-generation-on-vcgbench-1-0-test.md?metric=TU>): Higher is better; unit: SCALAR.

- [CO](<https://www.sota2.com/research/sota/video-based-text-generation-on-vcgbench-1-0-test.md?metric=CO>): Higher is better; unit: SCALAR.

- [Average Score](<https://www.sota2.com/research/sota/video-based-text-generation-on-vcgbench-1-0-test.md?metric=Average%20Score>): Higher is better; unit: SCALAR.

## Recorded results

| Method | CI | Result date | Report paper |
| --- | --- | --- | --- |
| LLaVA-NeXT-Video-DPO | 3.81 | Not recorded | 68fa74c91e209131cea63cb0 |
| LLaVA-NeXT-Video | 3.48 | Not recorded | 68fa74c91e209131cea63cb0 |
| SF-LLaVA-34B | 3.48 | Not recorded | 68fa74c91e209131cea63cb0 |
| LLaVA-NeXT-Image | 3.29 | Not recorded | 68fa74c91e209131cea63cb0 |
| IG-VLM (LLaVA-v1.6) | 3.11 | Not recorded | 68fa74c91e209131cea63cb0 |
| Video-LLaMA2 | 3.08 | Not recorded | 68fa74c91e209131cea63cb0 |

### LLaVA-NeXT-Video-DPO

CI: 3.81. Result ID: 69937b953a6c314849164307.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74c91e209131cea63cb0.md>)

Recorded evaluation configuration: {"LLM Size":"34B","Vision Encoder":"CLIP-L","Training Protocol":"SFT"}. This record does not assert independent reproduction.

### LLaVA-NeXT-Video

CI: 3.48. Result ID: 69937b953a6c314849164306.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74c91e209131cea63cb0.md>)

Recorded evaluation configuration: {"LLM Size":"34B","Vision Encoder":"CLIP-L","Training Protocol":"SFT"}. This record does not assert independent reproduction.

### SF-LLaVA-34B

CI: 3.48. Result ID: 69937b953a6c31484916430a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74c91e209131cea63cb0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74c91e209131cea63cb0.md>)

[Implementation](<https://huggingface.co/collections/liuhaotian/llava-next-65b9e40155f60fd046a5ccf2>)

Recorded evaluation configuration: {"LLM Size":"34B","Vision Encoder":"CLIP-L","Training Protocol":"Training-free"}. This record does not assert independent reproduction.

### LLaVA-NeXT-Image

CI: 3.29. Result ID: 69937b953a6c314849164308.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74c91e209131cea63cb0.md>)

Recorded evaluation configuration: {"LLM Size":"34B","Vision Encoder":"CLIP-L","Training Protocol":"Training-free"}. This record does not assert independent reproduction.

### IG-VLM (LLaVA-v1.6)

CI: 3.11. Result ID: 69937b953a6c314849164309.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74c91e209131cea63cb0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72b71e209131cea581e0.md>)

[Implementation](<https://github.com/imagegridworth/IG-VLM>)

Recorded evaluation configuration: {"LLM Size":"34B","Vision Encoder":"CLIP-L","Training Protocol":"Training-free"}. This record does not assert independent reproduction.

### Video-LLaMA2

CI: 3.08. Result ID: 69937b953a6c314849164305.

[Reporting paper](<https://www.sota2.com/research/paper/68fa74c91e209131cea63cb0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681dd.md>)

[Implementation](<https://github.com/DAMO-NLP-SG/VideoLLaMA2>)

Recorded evaluation configuration: {"LLM Size":"46.7B","Vision Encoder":"CLIP-L","Training Protocol":"SFT"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.