Multi-turn Math on GSM8K (multi-turn)
89.8Average AccuracyFIC
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| FICSetting=FULL, Backbone=PHI-4-14B2026.05 | 89.8 | — | — | |
| BaselineSetting=CONCAT, Backbone=PHI-4-14B2026.05 | 89.4 | — | — | |
| BaselineSetting=FULL, Backbone=PHI-4-14B2026.05 | 88.3 | — | — | |
| FICSetting=CONCAT, Backbone=PHI-4-14B2026.05 | 88 | — | — | |
| BaselineSetting=CONCAT, Backbone=QWEN2.5-7B2026.05 | 85 | — | — | |
| BaselineSetting=FULL, Backbone=QWEN2.5-7B2026.05 | 84.2 | — | — | |
| FICSetting=FULL, Backbone=QWEN2.5-7B2026.05 | 83.6 | — | — | |
| BaselineSetting=FULL, Backbone=OLMO-2-13B2026.05 | 82.9 | — | — | |
| FICSetting=CONCAT, Backbone=QWEN2.5-7B2026.05 | 82.6 | — | — | |
| FICSetting=SHARDED, Backbone=PHI-4-14B2026.05 | 82.4 | 35 | 92 | |
| BaselineSetting=CONCAT, Backbone=OLMO-2-13B2026.05 | 82.2 | — | — | |
| FICSetting=FULL, Backbone=OLMO-2-13B2026.05 | 79.5 | — | — | |
| FICSetting=SHARDED, Backbone=QWEN2.5-7B2026.05 | 78.7 | 34 | 93 | |
| FICSetting=CONCAT, Backbone=OLMO-2-13B2026.05 | 77.8 | — | — | |
| FICSetting=SHARDED, Backbone=OLMO-2-13B2026.05 | 77.1 | 42 | 94 | |
| FICSetting=FULL, Backbone=LLAMA-3.1-8B2026.05 | 73.8 | — | — | |
| BaselineSetting=FULL, Backbone=LLAMA-3.1-8B2026.05 | 72.5 | — | — | |
| FICSetting=CONCAT, Backbone=LLAMA-3.1-8B2026.05 | 72.3 | — | — | |
| FICSetting=SHARDED, Backbone=LLAMA-3.1-8B2026.05 | 71.7 | 60 | 100 | |
| BaselineSetting=CONCAT, Backbone=LLAMA-3.1-8B2026.05 | 71.5 | — | — | |
| BaselineSetting=FULL, Backbone=LLAMA-3.2-3B2026.05 | 62.5 | — | — | |
| BaselineSetting=SHARDED, Backbone=PHI-4-14B2026.05 | 61.2 | — | — | |
| FICSetting=SHARDED, Backbone=LLAMA-3.2-3B2026.05 | 60.4 | 54 | 100 | |
| BaselineSetting=CONCAT, Backbone=LLAMA-3.2-3B2026.05 | 60.2 | — | — | |
| BaselineSetting=SHARDED, Backbone=QWEN2.5-7B2026.05 | 58.9 | — | — | |
| FICSetting=FULL, Backbone=LLAMA-3.2-3B2026.05 | 58.3 | — | — | |
| BaselineSetting=SHARDED, Backbone=OLMO-2-13B2026.05 | 54.4 | — | — | |
| FICSetting=CONCAT, Backbone=LLAMA-3.2-3B2026.05 | 52.6 | — | — | |
| BaselineSetting=SHARDED, Backbone=LLAMA-3.1-8B2026.05 | 44.7 | — | — | |
| BaselineSetting=SHARDED, Backbone=LLAMA-3.2-3B2026.05 | 39.2 | — | — |