# Faithfulness Measurement on Alpaca Source: [Faithfulness Measurement on Alpaca]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:36:45.550Z Evaluation of model faithfulness using the Alpaca dataset, measured by metrics including BLEU, ROUGE-L, SentenceBert, PR, and KL divergence. 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 | Faithfulness Measurement | | Dataset (including any stated split) | Alpaca | | Metric | BLEU | | 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 - [BLEU](): Higher is better; unit: SCALAR. - [ROUGE-L Precision](): Higher is better; unit: SCALAR. - [ROUGE-L Recall](): Higher is better; unit: SCALAR. - [ROUGE-L F1](): Higher is better; unit: SCALAR. - [SentenceBert](): Higher is better; unit: SCALAR. - [PR](): Higher is better; unit: SCALAR. - [KL Divergence](): Lower is better; unit: SCALAR. ## Recorded results | Method | BLEU | Result date | Report paper | | --- | --- | --- | --- | | Random | 0.601 | Not recorded | 68fa7b651e209131cea882f8 | | Random | 0.587 | Not recorded | 68fa7b651e209131cea882f8 | | Attention | 0.547 | Not recorded | 68fa7b651e209131cea882f8 | | Integrated-Gradient | 0.541 | Not recorded | 68fa7b651e209131cea882f8 | | Integrated-Gradient | 0.541 | Not recorded | 68fa7b651e209131cea882f8 | | Last-Attention | 0.533 | Not recorded | 68fa7b651e209131cea882f8 | | Captum | 0.515 | Not recorded | 68fa7b651e209131cea882f8 | | Attention | 0.486 | Not recorded | 68fa7b651e209131cea882f8 | | JoPA | 0.484 | Not recorded | 68fa7b651e209131cea882f8 | | Last-Attention | 0.475 | Not recorded | 68fa7b651e209131cea882f8 | | Captum | 0.466 | Not recorded | 68fa7b651e209131cea882f8 | | JoPA | 0.433 | Not recorded | 68fa7b651e209131cea882f8 | ### Random BLEU: 0.601. Result ID: 698b852e7a967eddb76f42ca. [Reporting paper]() Recorded evaluation configuration: {"Model":"LlaMA-2 (7B-Chat)"}. This record does not assert independent reproduction. ### Random BLEU: 0.587. Result ID: 698b852e7a967eddb76f42d0. [Reporting paper]() Recorded evaluation configuration: {"Model":"Vicuna (7B)"}. This record does not assert independent reproduction. ### Attention BLEU: 0.547. Result ID: 698b852e7a967eddb76f42cc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LlaMA-2 (7B-Chat)"}. This record does not assert independent reproduction. ### Integrated-Gradient BLEU: 0.541. Result ID: 698b852e7a967eddb76f42ce. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LlaMA-2 (7B-Chat)"}. This record does not assert independent reproduction. ### Integrated-Gradient BLEU: 0.541. Result ID: 698b852e7a967eddb76f42d4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Vicuna (7B)"}. This record does not assert independent reproduction. ### Last-Attention BLEU: 0.533. Result ID: 698b852e7a967eddb76f42cb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LlaMA-2 (7B-Chat)"}. This record does not assert independent reproduction. ### Captum BLEU: 0.515. Result ID: 698b852e7a967eddb76f42cd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LlaMA-2 (7B-Chat)"}. This record does not assert independent reproduction. ### Attention BLEU: 0.486. Result ID: 698b852e7a967eddb76f42d2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Vicuna (7B)"}. This record does not assert independent reproduction. ### JoPA BLEU: 0.484. Result ID: 698b852e7a967eddb76f42cf. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model":"LlaMA-2 (7B-Chat)"}. This record does not assert independent reproduction. ### Last-Attention BLEU: 0.475. Result ID: 698b852e7a967eddb76f42d1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Vicuna (7B)"}. This record does not assert independent reproduction. ### Captum BLEU: 0.466. Result ID: 698b852e7a967eddb76f42d3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Vicuna (7B)"}. This record does not assert independent reproduction. ### JoPA BLEU: 0.433. Result ID: 698b852e7a967eddb76f42d5. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model":"Vicuna (7B)"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.