# One-step next-observation prediction on AgentGym Unweighted Average (test)

Source: [One-step next-observation prediction on AgentGym Unweighted Average (test)](<https://www.sota2.com/research/sota/one-step-next-observation-prediction-on-agentgym-unweighted-average-test>)

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

Data updated: 2026-06-01T08:26:47.894Z

Evaluation of One-step next-observation prediction performance on the AgentGym Unweighted Average (test) split, measured using Token F1 and BLEU-4.

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 | One-step next-observation prediction |
| Dataset (including any stated split) | AgentGym Unweighted Average (test) |
| Metric | Token F1 |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 16 of 16 |
| 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

- [Token F1](<https://www.sota2.com/research/sota/one-step-next-observation-prediction-on-agentgym-unweighted-average-test.md?metric=Token%20F1>): Higher is better; unit: PERCENT.

- [BLEU-4](<https://www.sota2.com/research/sota/one-step-next-observation-prediction-on-agentgym-unweighted-average-test.md?metric=BLEU-4>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Token F1 | Result date | Report paper |
| --- | --- | --- | --- |
| Word2World | 85 | Not recorded | 6a1d11dedbd7150506f83dbe |
| PatchWorld-Residual | 70 | Not recorded | 6a1d11dedbd7150506f83dbe |
| PatchWorld-Residual | 70 | Not recorded | 6a1d11dedbd7150506f83dbe |
| PatchWorld-Residual | 69 | Not recorded | 6a1d11dedbd7150506f83dbe |
| LLM-Direct | 66 | Not recorded | 6a1d11dedbd7150506f83dbe |
| PoE-World | 65 | Not recorded | 6a1d11dedbd7150506f83dbe |
| LLM-Direct | 64 | Not recorded | 6a1d11dedbd7150506f83dbe |
| LLM-Direct | 63 | Not recorded | 6a1d11dedbd7150506f83dbe |
| WorldCoder | 63 | Not recorded | 6a1d11dedbd7150506f83dbe |
| WorldCoder | 63 | Not recorded | 6a1d11dedbd7150506f83dbe |
| PoE-World | 63 | Not recorded | 6a1d11dedbd7150506f83dbe |
| PatchWorld-Simple | 60 | Not recorded | 6a1d11dedbd7150506f83dbe |
| PatchWorld-Simple | 60 | Not recorded | 6a1d11dedbd7150506f83dbe |
| WorldCoder | 57 | Not recorded | 6a1d11dedbd7150506f83dbe |
| PatchWorld-Simple | 57 | Not recorded | 6a1d11dedbd7150506f83dbe |
| PoE-World | 51 | Not recorded | 6a1d11dedbd7150506f83dbe |

### Word2World

Token F1: 85. Result ID: 6a1d4236fbc373592c166ac7.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/694a22409eebe296e72aa9f3.md>)

[Implementation](<https://github.com/X1AOX1A/Word2World>)

Recorded evaluation configuration: {"Backbone":"Qwen3.5-4B","LLM Usage":"SFT","Infer. LLM?":"Yes"}. This record does not assert independent reproduction.

### PatchWorld-Residual

Token F1: 70. Result ID: 6a1d4236fbc373592c166ad5.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

Recorded evaluation configuration: {"Backbone":"Mimo-v2.5","LLM Usage":"CodeGen","Infer. LLM?":"No"}. This record does not assert independent reproduction.

### PatchWorld-Residual

Token F1: 70. Result ID: 6a1d4236fbc373592c166ad6.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-V4-Flash","LLM Usage":"CodeGen","Infer. LLM?":"No"}. This record does not assert independent reproduction.

### PatchWorld-Residual

Token F1: 69. Result ID: 6a1d4236fbc373592c166ad4.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-Coder-480B","LLM Usage":"CodeGen","Infer. LLM?":"No"}. This record does not assert independent reproduction.

### LLM-Direct

Token F1: 66. Result ID: 6a1d4236fbc373592c166aca.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-V4-Flash","LLM Usage":"ICL","Infer. LLM?":"Yes"}. This record does not assert independent reproduction.

### PoE-World

Token F1: 65. Result ID: 6a1d4236fbc373592c166acf.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c301e209131cea8ceb8.md>)

Recorded evaluation configuration: {"Backbone":"Mimo-v2.5","LLM Usage":"CodeGen","Infer. LLM?":"No"}. This record does not assert independent reproduction.

### LLM-Direct

Token F1: 64. Result ID: 6a1d4236fbc373592c166ac8.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-Coder-480B","LLM Usage":"ICL","Infer. LLM?":"Yes"}. This record does not assert independent reproduction.

### LLM-Direct

Token F1: 63. Result ID: 6a1d4236fbc373592c166ac9.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

Recorded evaluation configuration: {"Backbone":"Mimo-v2.5","LLM Usage":"ICL","Infer. LLM?":"Yes"}. This record does not assert independent reproduction.

### WorldCoder

Token F1: 63. Result ID: 6a1d4236fbc373592c166acb.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74e71e209131cea64986.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-Coder-480B","LLM Usage":"CodeGen","Infer. LLM?":"No"}. This record does not assert independent reproduction.

### WorldCoder

Token F1: 63. Result ID: 6a1d4236fbc373592c166acc.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74e71e209131cea64986.md>)

Recorded evaluation configuration: {"Backbone":"Mimo-v2.5","LLM Usage":"CodeGen","Infer. LLM?":"No"}. This record does not assert independent reproduction.

### PoE-World

Token F1: 63. Result ID: 6a1d4236fbc373592c166ace.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c301e209131cea8ceb8.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-Coder-480B","LLM Usage":"CodeGen","Infer. LLM?":"No"}. This record does not assert independent reproduction.

### PatchWorld-Simple

Token F1: 60. Result ID: 6a1d4236fbc373592c166ad2.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

Recorded evaluation configuration: {"Backbone":"Mimo-v2.5","LLM Usage":"CodeGen","Infer. LLM?":"No"}. This record does not assert independent reproduction.

### PatchWorld-Simple

Token F1: 60. Result ID: 6a1d4236fbc373592c166ad3.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-V4-Flash","LLM Usage":"CodeGen","Infer. LLM?":"No"}. This record does not assert independent reproduction.

### WorldCoder

Token F1: 57. Result ID: 6a1d4236fbc373592c166acd.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74e71e209131cea64986.md>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-V4-Flash","LLM Usage":"CodeGen","Infer. LLM?":"No"}. This record does not assert independent reproduction.

### PatchWorld-Simple

Token F1: 57. Result ID: 6a1d4236fbc373592c166ad1.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-Coder-480B","LLM Usage":"CodeGen","Infer. LLM?":"No"}. This record does not assert independent reproduction.

### PoE-World

Token F1: 51. Result ID: 6a1d4236fbc373592c166ad0.

[Reporting paper](<https://www.sota2.com/research/paper/6a1d11dedbd7150506f83dbe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c301e209131cea8ceb8.md>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-V4-Flash","LLM Usage":"CodeGen","Infer. LLM?":"No"}. This record does not assert independent reproduction.

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