List your company

Mathematical Reasoning on GSM8K (Greedy, Pass@8)

66.9Greedy AccuracySFT
13.3427.24541.1555.055May 28, 2026

Evaluation results

MethodLinks
SFTModel=Llama-3.2-3B, Variant=w/ CoT, TTFT (ms)=754.42026.0566.978.3
RiMModel=Llama-3.2-3B, Variant=Any block, TTFT (ms)=27.92026.0557.371.8
RiMModel=Llama-3.2-1B, Variant=Any block, TTFT (ms)=16.12026.0551.476.8
CoconutBackbone=Llama-3.2-1B2026.0550.6—
RiMBackbone=Llama-3.2-3B2026.0549.4—
SFTModel=Llama-3.2-1B, Variant=w/ CoT, TTFT (ms)=420.32026.0549.164.7
RiMModel=Llama-3.2-3B, Variant=Final block, TTFT (ms)=27.92026.0548.858.8
DARTBackbone=Llama-3.2-3B2026.0546.6—
CoconutBackbone=Llama-3.2-3B2026.0543.6—
RiMBackbone=Llama-3.2-1B2026.0543.1—
DARTBackbone=Llama-3.2-1B2026.0542.6—
CoconutModel=Llama-3.2-3B, Variant=w/o Stage 0, TTFT (ms)=188.82026.0542.256.8
RiMModel=Llama-3.2-1B, Variant=Final block, TTFT (ms)=16.12026.0542.156.1
CoconutModel=Llama-3.2-3B, Variant=w/ Stage 0, TTFT (ms)=188.82026.0541.355.5
SFTModel=GPT-2, Variant=w/ CoT, TTFT (ms)=213.72026.0539.857
RiMModel=GPT-2, Variant=Any block, TTFT (ms)=7.62026.0539.578.1
CoconutModel=Llama-3.2-1B, Variant=w/ Stage 0, TTFT (ms)=108.32026.0536.951.1
SFTModel=Llama-3.2-3B, Variant=w/o CoT, TTFT (ms)=27.92026.0536.245.9
RiMBackbone=GPT-22026.0535.5—
CoconutBackbone=GPT-22026.0534.1—
RiMModel=GPT-2, Variant=Final block, TTFT (ms)=7.62026.0533.649.1
CoconutModel=GPT-2, Variant=w/ Stage 0, TTFT (ms)=53.42026.0531.145
CoconutModel=Llama-3.2-1B, Variant=w/o Stage 0, TTFT (ms)=108.32026.0530.146.1
DARTBackbone=GPT-22026.0524.7—
SFTModel=Llama-3.2-1B, Variant=w/o CoT, TTFT (ms)=16.12026.0523.941.7
CoconutModel=GPT-2, Variant=w/o Stage 0, TTFT (ms)=53.42026.0522.340.6
SFTModel=GPT-2, Variant=w/o CoT, TTFT (ms)=7.62026.0515.433.3