# Circuit Recovery Evaluation Suite (IOI, MCQA, Arithmetic, ARC) Llama-3 (3B to 8B) Transfer (test)

Source: [Circuit Recovery Evaluation Suite (IOI, MCQA, Arithmetic, ARC) Llama-3 (3B to 8B) Transfer (test)](<https://www.sota2.com/research/sota/circuit-recovery-evaluation-suite-ioi-mcqa-arithmetic-arc-llama-3-3b-to-8b-transfer-test>)

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

Data updated: 2026-04-28T10:11:43.768Z

Evaluation of model performance on the Circuit Recovery task using the specific test split of the Circuit Recovery Evaluation Suite, focusing on Llama-3 models ranging from 3B to 8B parameters across IOI, MCQA, Arithmetic, and ARC components.

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 | Circuit Recovery |
| Dataset (including any stated split) | Circuit Recovery Evaluation Suite (IOI, MCQA, Arithmetic, ARC) Llama-3 (3B to 8B) Transfer (test) |
| Metric | Arithmetic (+) Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 13 of 13 |
| 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

- [Arithmetic (+) Score](<https://www.sota2.com/research/sota/circuit-recovery-evaluation-suite-ioi-mcqa-arithmetic-arc-llama-3-3b-to-8b-transfer-test.md?metric=Arithmetic%20(%2B)%20Score>): Higher is better; unit: PERCENT.

- [IOI Score](<https://www.sota2.com/research/sota/circuit-recovery-evaluation-suite-ioi-mcqa-arithmetic-arc-llama-3-3b-to-8b-transfer-test.md?metric=IOI%20Score>): Higher is better; unit: PERCENT.

- [MCQA Score](<https://www.sota2.com/research/sota/circuit-recovery-evaluation-suite-ioi-mcqa-arithmetic-arc-llama-3-3b-to-8b-transfer-test.md?metric=MCQA%20Score>): Higher is better; unit: PERCENT.

- [Arithmetic (-) Score](<https://www.sota2.com/research/sota/circuit-recovery-evaluation-suite-ioi-mcqa-arithmetic-arc-llama-3-3b-to-8b-transfer-test.md?metric=Arithmetic%20(-)%20Score>): Higher is better; unit: PERCENT.

- [ARC-E Score](<https://www.sota2.com/research/sota/circuit-recovery-evaluation-suite-ioi-mcqa-arithmetic-arc-llama-3-3b-to-8b-transfer-test.md?metric=ARC-E%20Score>): Higher is better; unit: PERCENT.

- [ARC-C Score](<https://www.sota2.com/research/sota/circuit-recovery-evaluation-suite-ioi-mcqa-arithmetic-arc-llama-3-3b-to-8b-transfer-test.md?metric=ARC-C%20Score>): Higher is better; unit: PERCENT.

- [Average Score](<https://www.sota2.com/research/sota/circuit-recovery-evaluation-suite-ioi-mcqa-arithmetic-arc-llama-3-3b-to-8b-transfer-test.md?metric=Average%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Arithmetic (+) Score | Result date | Report paper |
| --- | --- | --- | --- |
| NAP-IG-Inputs | 46 | Not recorded | 69f03f7555a3c32af89fa4df |
| NAP-IG-Activations | 46 | Not recorded | 69f03f7555a3c32af89fa4df |
| NAP-IG-Inputs | 45 | Not recorded | 69f03f7555a3c32af89fa4df |
| NAP | 40 | Not recorded | 69f03f7555a3c32af89fa4df |
| NAP | 36 | Not recorded | 69f03f7555a3c32af89fa4df |
| NAP-IG-Activations | 33 | Not recorded | 69f03f7555a3c32af89fa4df |
| NAP-IG-Inputs | 29 | Not recorded | 69f03f7555a3c32af89fa4df |
| NAP-IG-Activations | 28 | Not recorded | 69f03f7555a3c32af89fa4df |
| NAP-IG-Inputs | 27 | Not recorded | 69f03f7555a3c32af89fa4df |
| NAP | 27 | Not recorded | 69f03f7555a3c32af89fa4df |
| NAP-IG-Activations | 27 | Not recorded | 69f03f7555a3c32af89fa4df |
| Baseline | 25 | Not recorded | 69f03f7555a3c32af89fa4df |
| NAP | 25 | Not recorded | 69f03f7555a3c32af89fa4df |

### NAP-IG-Inputs

Arithmetic (+) Score: 46. Result ID: 69f087cc47f34ce924d70f27.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

Recorded evaluation configuration: {"Method Category":"NAP-IG-Inputs","Transfer Setting":"Gold (Upper Bound)","Transfer Direction":"Llama-3 (3B → 8B)"}. This record does not assert independent reproduction.

### NAP-IG-Activations

Arithmetic (+) Score: 46. Result ID: 69f087cc47f34ce924d70f30.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

Recorded evaluation configuration: {"Method Category":"NAP-IG-Activations","Transfer Setting":"DFA (In-Distribution)","Transfer Direction":"Llama-3 (3B → 8B)"}. This record does not assert independent reproduction.

### NAP-IG-Inputs

Arithmetic (+) Score: 45. Result ID: 69f087cc47f34ce924d70f28.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

Recorded evaluation configuration: {"Method Category":"NAP-IG-Inputs","Transfer Setting":"DFA (In-Distribution)","Transfer Direction":"Llama-3 (3B → 8B)"}. This record does not assert independent reproduction.

### NAP

Arithmetic (+) Score: 40. Result ID: 69f087cc47f34ce924d70f2c.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

Recorded evaluation configuration: {"Method Category":"NAP","Transfer Setting":"DFA (In-Distribution)","Transfer Direction":"Llama-3 (3B → 8B)"}. This record does not assert independent reproduction.

### NAP

Arithmetic (+) Score: 36. Result ID: 69f087cc47f34ce924d70f2d.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

Recorded evaluation configuration: {"Method Category":"NAP","Transfer Setting":"DFA (Near-Distribution)","Transfer Direction":"Llama-3 (3B → 8B)"}. This record does not assert independent reproduction.

### NAP-IG-Activations

Arithmetic (+) Score: 33. Result ID: 69f087cc47f34ce924d70f2f.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

Recorded evaluation configuration: {"Method Category":"NAP-IG-Activations","Transfer Setting":"Gold (Upper Bound)","Transfer Direction":"Llama-3 (3B → 8B)"}. This record does not assert independent reproduction.

### NAP-IG-Inputs

Arithmetic (+) Score: 29. Result ID: 69f087cc47f34ce924d70f29.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

Recorded evaluation configuration: {"Method Category":"NAP-IG-Inputs","Transfer Setting":"DFA (Near-Distribution)","Transfer Direction":"Llama-3 (3B → 8B)"}. This record does not assert independent reproduction.

### NAP-IG-Activations

Arithmetic (+) Score: 28. Result ID: 69f087cc47f34ce924d70f32.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

Recorded evaluation configuration: {"Method Category":"NAP-IG-Activations","Transfer Setting":"DFA (Far/Zero-Shot)","Transfer Direction":"Llama-3 (3B → 8B)"}. This record does not assert independent reproduction.

### NAP-IG-Inputs

Arithmetic (+) Score: 27. Result ID: 69f087cc47f34ce924d70f2a.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

Recorded evaluation configuration: {"Method Category":"NAP-IG-Inputs","Transfer Setting":"DFA (Far/Zero-Shot)","Transfer Direction":"Llama-3 (3B → 8B)"}. This record does not assert independent reproduction.

### NAP

Arithmetic (+) Score: 27. Result ID: 69f087cc47f34ce924d70f2e.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

Recorded evaluation configuration: {"Method Category":"NAP","Transfer Setting":"DFA (Far/Zero-Shot)","Transfer Direction":"Llama-3 (3B → 8B)"}. This record does not assert independent reproduction.

### NAP-IG-Activations

Arithmetic (+) Score: 27. Result ID: 69f087cc47f34ce924d70f31.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

Recorded evaluation configuration: {"Method Category":"NAP-IG-Activations","Transfer Setting":"DFA (Near-Distribution)","Transfer Direction":"Llama-3 (3B → 8B)"}. This record does not assert independent reproduction.

### Baseline

Arithmetic (+) Score: 25. Result ID: 69f087cc47f34ce924d70f26.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

Recorded evaluation configuration: {"Method Category":"Baseline","Transfer Setting":"Random Alignment","Transfer Direction":"Llama-3 (3B → 8B)"}. This record does not assert independent reproduction.

### NAP

Arithmetic (+) Score: 25. Result ID: 69f087cc47f34ce924d70f2b.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f7555a3c32af89fa4df.md>)

Recorded evaluation configuration: {"Method Category":"NAP","Transfer Setting":"Gold (Upper Bound)","Transfer Direction":"Llama-3 (3B → 8B)"}. This record does not assert independent reproduction.

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