# Dialogue Generation on NEGOTIATIONTOM N=100

Source: [Dialogue Generation on NEGOTIATIONTOM N=100](<https://www.sota2.com/research/sota/dialogue-generation-on-negotiationtom-n-100>)

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

Data updated: 2026-04-14T07:50:44.296Z

Evaluation of Dialogue Generation performance on the NEGOTIATIONTOM N=100 dataset, assessed using Theory of Mind (ToM), Coherence (Coh.), and NSE metrics based on evaluations from Llama-3.3-70B, GPT-5.1, and Human judges.

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 | Dialogue Generation |
| Dataset (including any stated split) | NEGOTIATIONTOM N=100 |
| Metric | ToM (Llama-3.3-70B) |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 12 of 12 |
| 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

- [ToM (Llama-3.3-70B)](<https://www.sota2.com/research/sota/dialogue-generation-on-negotiationtom-n-100.md?metric=ToM%20(Llama-3.3-70B)>): Higher is better; unit: SCALAR.

- [Coherence (Llama-3.3-70B)](<https://www.sota2.com/research/sota/dialogue-generation-on-negotiationtom-n-100.md?metric=Coherence%20(Llama-3.3-70B)>): Higher is better; unit: SCALAR.

- [NSE (Llama-3.3-70B)](<https://www.sota2.com/research/sota/dialogue-generation-on-negotiationtom-n-100.md?metric=NSE%20(Llama-3.3-70B)>): Higher is better; unit: SCALAR.

- [ToM (GPT-5.1)](<https://www.sota2.com/research/sota/dialogue-generation-on-negotiationtom-n-100.md?metric=ToM%20(GPT-5.1)>): Higher is better; unit: SCALAR.

- [Coherence (GPT-5.1)](<https://www.sota2.com/research/sota/dialogue-generation-on-negotiationtom-n-100.md?metric=Coherence%20(GPT-5.1)>): Higher is better; unit: SCALAR.

- [NSE (GPT-5.1)](<https://www.sota2.com/research/sota/dialogue-generation-on-negotiationtom-n-100.md?metric=NSE%20(GPT-5.1)>): Higher is better; unit: SCALAR.

- [ToM (Human)](<https://www.sota2.com/research/sota/dialogue-generation-on-negotiationtom-n-100.md?metric=ToM%20(Human)>): Higher is better; unit: SCALAR.

- [Coherence (Human)](<https://www.sota2.com/research/sota/dialogue-generation-on-negotiationtom-n-100.md?metric=Coherence%20(Human)>): Higher is better; unit: SCALAR.

- [NSE (Human)](<https://www.sota2.com/research/sota/dialogue-generation-on-negotiationtom-n-100.md?metric=NSE%20(Human)>): Higher is better; unit: SCALAR.

## Recorded results

| Method | ToM (Llama-3.3-70B) | Result date | Report paper |
| --- | --- | --- | --- |
| COSTOM | 0.751 | Not recorded | 69ddc9ed7d0172140ec79860 |
| COSTOM | 0.499 | Not recorded | 69ddc9ed7d0172140ec79860 |
| LatentQA | 0.44 | Not recorded | 69ddc9ed7d0172140ec79860 |
| MindDial (Fine-tuned) | 0.405 | Not recorded | 69ddc9ed7d0172140ec79860 |
| MindDial (prompt) | 0.306 | Not recorded | 69ddc9ed7d0172140ec79860 |
| Full-Layer LoRA | 0.245 | Not recorded | 69ddc9ed7d0172140ec79860 |
| LatentQA | 0.221 | Not recorded | 69ddc9ed7d0172140ec79860 |
| Full-Layer LoRA | 0.153 | Not recorded | 69ddc9ed7d0172140ec79860 |
| MindDial (Fine-tuned) | 0.149 | Not recorded | 69ddc9ed7d0172140ec79860 |
| MindDial (prompt) | 0.118 | Not recorded | 69ddc9ed7d0172140ec79860 |
| Zero-shot Baseline | 0.081 | Not recorded | 69ddc9ed7d0172140ec79860 |
| Zero-shot Baseline | 0.017 | Not recorded | 69ddc9ed7d0172140ec79860 |

### COSTOM

ToM (Llama-3.3-70B): 0.751. Result ID: 69ddf18d47f34ce924c51d19.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9ed7d0172140ec79860.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddc9ed7d0172140ec79860.md>)

Recorded evaluation configuration: {"Base Model":"Qwen2.5-7B-Instruct","Method Type":"Tuning & Intervention"}. This record does not assert independent reproduction.

### COSTOM

ToM (Llama-3.3-70B): 0.499. Result ID: 69ddf18d47f34ce924c51d13.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9ed7d0172140ec79860.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddc9ed7d0172140ec79860.md>)

Recorded evaluation configuration: {"Base Model":"Llama-3-8B-Instruct","Method Type":"Tuning & Intervention"}. This record does not assert independent reproduction.

### LatentQA

ToM (Llama-3.3-70B): 0.44. Result ID: 69ddf18d47f34ce924c51d18.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9ed7d0172140ec79860.md>)

Recorded evaluation configuration: {"Base Model":"Qwen2.5-7B-Instruct","Method Type":"Tuning & Intervention"}. This record does not assert independent reproduction.

### MindDial (Fine-tuned)

ToM (Llama-3.3-70B): 0.405. Result ID: 69ddf18d47f34ce924c51d10.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9ed7d0172140ec79860.md>)

Recorded evaluation configuration: {"Base Model":"Llama-3-8B-Instruct","Method Type":"Tuning & Intervention"}. This record does not assert independent reproduction.

### MindDial (prompt)

ToM (Llama-3.3-70B): 0.306. Result ID: 69ddf18d47f34ce924c51d0f.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9ed7d0172140ec79860.md>)

Recorded evaluation configuration: {"Base Model":"Llama-3-8B-Instruct","Method Type":"Prompting Baselines"}. This record does not assert independent reproduction.

### Full-Layer LoRA

ToM (Llama-3.3-70B): 0.245. Result ID: 69ddf18d47f34ce924c51d11.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9ed7d0172140ec79860.md>)

Recorded evaluation configuration: {"Base Model":"Llama-3-8B-Instruct","Method Type":"Tuning & Intervention"}. This record does not assert independent reproduction.

### LatentQA

ToM (Llama-3.3-70B): 0.221. Result ID: 69ddf18d47f34ce924c51d12.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9ed7d0172140ec79860.md>)

Recorded evaluation configuration: {"Base Model":"Llama-3-8B-Instruct","Method Type":"Tuning & Intervention"}. This record does not assert independent reproduction.

### Full-Layer LoRA

ToM (Llama-3.3-70B): 0.153. Result ID: 69ddf18d47f34ce924c51d17.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9ed7d0172140ec79860.md>)

Recorded evaluation configuration: {"Base Model":"Qwen2.5-7B-Instruct","Method Type":"Tuning & Intervention"}. This record does not assert independent reproduction.

### MindDial (Fine-tuned)

ToM (Llama-3.3-70B): 0.149. Result ID: 69ddf18d47f34ce924c51d16.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9ed7d0172140ec79860.md>)

Recorded evaluation configuration: {"Base Model":"Qwen2.5-7B-Instruct","Method Type":"Tuning & Intervention"}. This record does not assert independent reproduction.

### MindDial (prompt)

ToM (Llama-3.3-70B): 0.118. Result ID: 69ddf18d47f34ce924c51d15.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9ed7d0172140ec79860.md>)

Recorded evaluation configuration: {"Base Model":"Qwen2.5-7B-Instruct","Method Type":"Prompting Baselines"}. This record does not assert independent reproduction.

### Zero-shot Baseline

ToM (Llama-3.3-70B): 0.081. Result ID: 69ddf18d47f34ce924c51d0e.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9ed7d0172140ec79860.md>)

Recorded evaluation configuration: {"Base Model":"Llama-3-8B-Instruct","Method Type":"Prompting Baselines"}. This record does not assert independent reproduction.

### Zero-shot Baseline

ToM (Llama-3.3-70B): 0.017. Result ID: 69ddf18d47f34ce924c51d14.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9ed7d0172140ec79860.md>)

Recorded evaluation configuration: {"Base Model":"Qwen2.5-7B-Instruct","Method Type":"Prompting Baselines"}. This record does not assert independent reproduction.

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