# Backward Transition Handling on PFDial-H (test)

Source: [Backward Transition Handling on PFDial-H (test)](<https://www.sota2.com/research/sota/backward-transition-handling-on-pfdial-h-test>)

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

Data updated: 2026-02-26T05:30:38.739Z

Evaluation of model performance on the Backward Transition Handling task using the test split of the PFDial-H dataset.

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 | Backward Transition Handling |
| Dataset (including any stated split) | PFDial-H (test) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 15 of 15 |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/backward-transition-handling-on-pfdial-h-test.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Backward Accuracy (Dist \< 5)](<https://www.sota2.com/research/sota/backward-transition-handling-on-pfdial-h-test.md?metric=Backward%20Accuracy%20(Dist%20%3C%205)>): Higher is better; unit: PERCENT.

- [Backward Accuracy (Dist \>= 5)](<https://www.sota2.com/research/sota/backward-transition-handling-on-pfdial-h-test.md?metric=Backward%20Accuracy%20(Dist%20%3E%3D%205)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Qwen2.5-7B | 76.25 | Not recorded | 68fa79591e209131cea7d332 |
| Qwen2.5-3B | 75 | Not recorded | 68fa79591e209131cea7d332 |
| Qwen2.5-1.5B | 70 | Not recorded | 68fa79591e209131cea7d332 |
| Qwen2.5-7B | 66.25 | Not recorded | 68fa79591e209131cea7d332 |
| Qwen2.5-0.5B | 65 | Not recorded | 68fa79591e209131cea7d332 |
| GPT-4o | 58.75 | Not recorded | 68fa79591e209131cea7d332 |
| Qwen2.5-0.5B | 58.75 | Not recorded | 68fa79591e209131cea7d332 |
| Claude-3.5-Sonnet | 57.5 | Not recorded | 68fa79591e209131cea7d332 |
| Qwen2.5-3B | 57.5 | Not recorded | 68fa79591e209131cea7d332 |
| GPT-3.5-Turbo | 51.25 | Not recorded | 68fa79591e209131cea7d332 |
| Qwen2.5-1.5B | 47.5 | Not recorded | 68fa79591e209131cea7d332 |
| DeepSeek-v3 | 40 | Not recorded | 68fa79591e209131cea7d332 |
| Qwen2.5-7B-Instruct | 31.25 | Not recorded | 68fa79591e209131cea7d332 |
| Gemini-2.0-Flash-Exp | 26.25 | Not recorded | 68fa79591e209131cea7d332 |
| LLaMA-3.1-8B-Instruct | 15 | Not recorded | 68fa79591e209131cea7d332 |

### Qwen2.5-7B

Accuracy: 76.25. Result ID: 6995db603badfaffe6e1e53d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Implementation](<https://github.com/KongLongGeFDU/PFDial>)

Recorded evaluation configuration: {"Training Approach":"Integrated Training"}. This record does not assert independent reproduction.

### Qwen2.5-3B

Accuracy: 75. Result ID: 6995db603badfaffe6e1e53c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Implementation](<https://github.com/KongLongGeFDU/PFDial>)

Recorded evaluation configuration: {"Training Approach":"Integrated Training"}. This record does not assert independent reproduction.

### Qwen2.5-1.5B

Accuracy: 70. Result ID: 6995db603badfaffe6e1e53b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Implementation](<https://github.com/KongLongGeFDU/PFDial>)

Recorded evaluation configuration: {"Training Approach":"Integrated Training"}. This record does not assert independent reproduction.

### Qwen2.5-7B

Accuracy: 66.25. Result ID: 6995db603badfaffe6e1e539.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Implementation](<https://github.com/KongLongGeFDU/PFDial>)

Recorded evaluation configuration: {"Training Approach":"Secondary Fine-tuning"}. This record does not assert independent reproduction.

### Qwen2.5-0.5B

Accuracy: 65. Result ID: 6995db603badfaffe6e1e53a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Implementation](<https://github.com/KongLongGeFDU/PFDial>)

Recorded evaluation configuration: {"Training Approach":"Integrated Training"}. This record does not assert independent reproduction.

### GPT-4o

Accuracy: 58.75. Result ID: 6995db603badfaffe6e1e535.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa727a1e209131cea56a1e.md>)

[Implementation](<https://github.com/openai/evals>)

Recorded evaluation configuration: {"Training Approach":"Baselines"}. This record does not assert independent reproduction.

### Qwen2.5-0.5B

Accuracy: 58.75. Result ID: 6995db603badfaffe6e1e536.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Implementation](<https://github.com/KongLongGeFDU/PFDial>)

Recorded evaluation configuration: {"Training Approach":"Secondary Fine-tuning"}. This record does not assert independent reproduction.

### Claude-3.5-Sonnet

Accuracy: 57.5. Result ID: 6995db603badfaffe6e1e530.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

Recorded evaluation configuration: {"Training Approach":"Baselines"}. This record does not assert independent reproduction.

### Qwen2.5-3B

Accuracy: 57.5. Result ID: 6995db603badfaffe6e1e538.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Implementation](<https://github.com/KongLongGeFDU/PFDial>)

Recorded evaluation configuration: {"Training Approach":"Secondary Fine-tuning"}. This record does not assert independent reproduction.

### GPT-3.5-Turbo

Accuracy: 51.25. Result ID: 6995db603badfaffe6e1e534.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6bf71e209131cea314c0.md>)

[Implementation](<https://github.com/openai/following-instructions-human-feedback>)

Recorded evaluation configuration: {"Training Approach":"Baselines"}. This record does not assert independent reproduction.

### Qwen2.5-1.5B

Accuracy: 47.5. Result ID: 6995db603badfaffe6e1e537.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Implementation](<https://github.com/KongLongGeFDU/PFDial>)

Recorded evaluation configuration: {"Training Approach":"Secondary Fine-tuning"}. This record does not assert independent reproduction.

### DeepSeek-v3

Accuracy: 40. Result ID: 6995db603badfaffe6e1e533.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77001e209131cea7066a.md>)

Recorded evaluation configuration: {"Training Approach":"Baselines"}. This record does not assert independent reproduction.

### Qwen2.5-7B-Instruct

Accuracy: 31.25. Result ID: 6995db603badfaffe6e1e531.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa767e1e209131cea6d6a7.md>)

Recorded evaluation configuration: {"Training Approach":"Baselines"}. This record does not assert independent reproduction.

### Gemini-2.0-Flash-Exp

Accuracy: 26.25. Result ID: 6995db603badfaffe6e1e532.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78731e209131cea78622.md>)

Recorded evaluation configuration: {"Training Approach":"Baselines"}. This record does not assert independent reproduction.

### LLaMA-3.1-8B-Instruct

Accuracy: 15. Result ID: 6995db603badfaffe6e1e52f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79591e209131cea7d332.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75d11e209131cea6a028.md>)

[Implementation](<https://github.com/meta-llama/llama-agentic-system>)

Recorded evaluation configuration: {"Training Approach":"Baselines"}. This record does not assert independent reproduction.

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