List your company

Multi-turn Math on GSM8K (multi-turn)

89.8Average AccuracyFIC
37.17650.83864.578.162May 23, 2026

Evaluation results

MethodLinks
FICSetting=FULL, Backbone=PHI-4-14B2026.0589.8——
BaselineSetting=CONCAT, Backbone=PHI-4-14B2026.0589.4——
BaselineSetting=FULL, Backbone=PHI-4-14B2026.0588.3——
FICSetting=CONCAT, Backbone=PHI-4-14B2026.0588——
BaselineSetting=CONCAT, Backbone=QWEN2.5-7B2026.0585——
BaselineSetting=FULL, Backbone=QWEN2.5-7B2026.0584.2——
FICSetting=FULL, Backbone=QWEN2.5-7B2026.0583.6——
BaselineSetting=FULL, Backbone=OLMO-2-13B2026.0582.9——
FICSetting=CONCAT, Backbone=QWEN2.5-7B2026.0582.6——
FICSetting=SHARDED, Backbone=PHI-4-14B2026.0582.43592
BaselineSetting=CONCAT, Backbone=OLMO-2-13B2026.0582.2——
FICSetting=FULL, Backbone=OLMO-2-13B2026.0579.5——
FICSetting=SHARDED, Backbone=QWEN2.5-7B2026.0578.73493
FICSetting=CONCAT, Backbone=OLMO-2-13B2026.0577.8——
FICSetting=SHARDED, Backbone=OLMO-2-13B2026.0577.14294
FICSetting=FULL, Backbone=LLAMA-3.1-8B2026.0573.8——
BaselineSetting=FULL, Backbone=LLAMA-3.1-8B2026.0572.5——
FICSetting=CONCAT, Backbone=LLAMA-3.1-8B2026.0572.3——
FICSetting=SHARDED, Backbone=LLAMA-3.1-8B2026.0571.760100
BaselineSetting=CONCAT, Backbone=LLAMA-3.1-8B2026.0571.5——
BaselineSetting=FULL, Backbone=LLAMA-3.2-3B2026.0562.5——
BaselineSetting=SHARDED, Backbone=PHI-4-14B2026.0561.2——
FICSetting=SHARDED, Backbone=LLAMA-3.2-3B2026.0560.454100
BaselineSetting=CONCAT, Backbone=LLAMA-3.2-3B2026.0560.2——
BaselineSetting=SHARDED, Backbone=QWEN2.5-7B2026.0558.9——
FICSetting=FULL, Backbone=LLAMA-3.2-3B2026.0558.3——
BaselineSetting=SHARDED, Backbone=OLMO-2-13B2026.0554.4——
FICSetting=CONCAT, Backbone=LLAMA-3.2-3B2026.0552.6——
BaselineSetting=SHARDED, Backbone=LLAMA-3.1-8B2026.0544.7——
BaselineSetting=SHARDED, Backbone=LLAMA-3.2-3B2026.0539.2——