List your company

Math on GSM8K (test)

94.04Mean@4TextGrad-M
-0.641623.939248.5273.1008May 31, 2025Jul 17, 2025Sep 3, 2025Oct 21, 2025Dec 7, 2025Jan 24, 2026Mar 13, 2026

Evaluation results

MethodLinks
TextGrad-MOptimization Strategy=Promptwise, LMforward=GPT-4o-mini, LMbackward=GPT-4o2025.0594.04—74.121.73
TextGrad-MOptimization Strategy=Blockwise, LMforward=GPT-4o-mini, LMbackward=GPT-4o2025.0593.98—74.592.2
TextGradValidation Protocol=w/ val revert, LMforward=GPT-4o-mini, LMbackward=GPT-4o2025.0593.74—72.39—
TextGradValidation Protocol=w/o val revert, LMforward=GPT-4o-mini, LMbackward=GPT-4o2025.0593.15—72.96—
GEPALMforward=GPT-4o-mini, LMbackward=GPT-4o2025.0591.55—72.11—
ReSyn#Params=7B, zero-shot=true, temperature=0.8, top-p=0.952026.0291.4———
AdalFlow-MOptimization Strategy=Promptwise, LMforward=GPT-4o-mini, LMbackward=GPT-4o2025.0590.78—72.630.8
AdalFlowLMforward=GPT-4o-mini, LMbackward=GPT-4o2025.0589—71.83—
COPRO-MOptimization Strategy=Promptwise, LMforward=GPT-4o-mini, LMbackward=GPT-4o2025.0588.74—68.790.94
COPROLMforward=GPT-4o-mini, LMbackward=GPT-4o2025.0587.99—67.84—
COPRO-MOptimization Strategy=Blockwise, LMforward=GPT-4o-mini, LMbackward=GPT-4o2025.0586.9—68.91.05
Qwen-2.5-Instruct#Params=7B, zero-shot=true, temperature=0.8, top-p=0.952026.0282.3———
Llama-3.1-Instruct#Params=8B, zero-shot=true, temperature=0.8, top-p=0.952026.0270.7———
MoBERatio=30%, # Params=0.96B2026.0354.353.8——
MoBERatio=40%, # Params=0.96B2026.0350.949——
MC-SMoERatio=30%, # Params=0.96B2026.0348.449.3——
MC-SMoERatio=40%, # Params=0.96B2026.0347.648.4——
HC-SMoERatio=40%, # Params=0.96B2026.034348.9——
MC-SMoERatio=50%, # Params=0.96B2026.0342.240.4——
MoBERatio=50%, # Params=0.96B2026.0339.338.6——
LoRARatio=0%, # Params=0.96B2026.0338.952——
HC-SMoERatio=30%, # Params=0.96B2026.0336.849.8——
HC-SMoERatio=50%, # Params=0.96B2026.0334.345.8——
LightMoERatio=40%, # Params=0.96B2026.0333.750.9——
LightMoERatio=30%, # Params=0.96B2026.0333.351.4——
LightMoERatio=50%, # Params=0.96B2026.0332.848.4——
Mistral-Instruct#Params=7B, zero-shot=true, temperature=0.8, top-p=0.952026.0222.3———
OriginalRatio=0%, # Params=-2026.03326.2——