# LLM Alignment on WikiText-2 raw v1 (train)

Source: [LLM Alignment on WikiText-2 raw v1 (train)](<https://www.sota2.com/research/sota/llm-alignment-on-wikitext-2-raw-v1-train>)

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

Data updated: 2026-04-27T06:29:26.007Z

Evaluation of LLM alignment capabilities using the training split of the WikiText-2 raw version 1 dataset, focusing on loss and time metrics related to proof generation and verification.

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 | LLM alignment |
| Dataset (including any stated split) | WikiText-2 raw v1 (train) |
| Metric | Loss |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 3 of 3 |
| 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

- [Loss](<https://www.sota2.com/research/sota/llm-alignment-on-wikitext-2-raw-v1-train.md?metric=Loss>): Lower is better; unit: SCALAR.

- [Training Time (min)](<https://www.sota2.com/research/sota/llm-alignment-on-wikitext-2-raw-v1-train.md?metric=Training%20Time%20(min)>): Lower is better; unit: TIME.

- [Proof Generation Time (min)](<https://www.sota2.com/research/sota/llm-alignment-on-wikitext-2-raw-v1-train.md?metric=Proof%20Generation%20Time%20(min)>): Lower is better; unit: TIME.

- [Verification Time (min)](<https://www.sota2.com/research/sota/llm-alignment-on-wikitext-2-raw-v1-train.md?metric=Verification%20Time%20(min)>): Lower is better; unit: TIME.

## Recorded results

| Method | Loss | Result date | Report paper |
| --- | --- | --- | --- |
| PoLO | 3.5737 | Not recorded | 68fa789a1e209131cea793d4 |
| PoLO | 3.5765 | Not recorded | 68fa789a1e209131cea793d4 |
| GD-PoL | 3.5893 | Not recorded | 68fa789a1e209131cea793d4 |

### PoLO

Loss: 3.5737. Result ID: 69ef011f47f34ce924d47310.

[Reporting paper](<https://www.sota2.com/research/paper/68fa789a1e209131cea793d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa789a1e209131cea793d4.md>)

Recorded evaluation configuration: {"Base Model":"DialoGPT-medium-345M","Watermark size (bit)":"512","\|S\| (%)":"27.27"}. This record does not assert independent reproduction.

### PoLO

Loss: 3.5765. Result ID: 69ef011f47f34ce924d47311.

[Reporting paper](<https://www.sota2.com/research/paper/68fa789a1e209131cea793d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa789a1e209131cea793d4.md>)

Recorded evaluation configuration: {"Base Model":"DialoGPT-medium-345M","Watermark size (bit)":"1024","\|S\| (%)":"25.00"}. This record does not assert independent reproduction.

### GD-PoL

Loss: 3.5893. Result ID: 69ef011f47f34ce924d4730f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa789a1e209131cea793d4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa697d1e209131cea22657.md>)

Recorded evaluation configuration: {"Base Model":"DialoGPT-medium-345M"}. This record does not assert independent reproduction.

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