List your company

Text Generation on LM1B (test)

2.46EntropyUDLM
2.3755362.9456683.51584.085932Dec 23, 2025Jan 19, 2026Feb 16, 2026Mar 16, 2026Apr 13, 2026May 11, 2026Jun 8, 2026

Evaluation results

MethodLinks
UDLMSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=42025.122.46—————————
PAIRFLOWSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=42025.122.48—————————
UDLMSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=82025.122.49—————————
PAIRFLOWSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=82025.122.49—————————
UDLMSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=162025.122.5—————————
UDLMSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=322025.122.5—————————
UDLMSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=642025.122.5—————————
UDLMSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=2562025.122.5—————————
UDLMSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=10242025.122.5—————————
PAIRFLOWSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=162025.122.51—————————
UDLMSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=1282025.122.51—————————
UDLMSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=5122025.122.51—————————
PAIRFLOWSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=322025.122.52—————————
PAIRFLOWSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=642025.122.52—————————
PAIRFLOWSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=2562025.122.52—————————
PAIRFLOWSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=10242025.122.52—————————
PAIRFLOWSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=1282025.122.53—————————
PAIRFLOWSequence Length (N)=16, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=5122025.122.53—————————
ILMEvaluator Model=Llama 3.2 3B2026.062.8————4.6721———
DILM-SEvaluator Model=Llama 3.2 3B2026.062.87————4.6530———
DILM-MEvaluator Model=Llama 3.2 3B2026.062.99————4.7648———
UDLMSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=42025.123.05—————————
PAIRFLOWSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=42025.123.06—————————
UDLMSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=82025.123.09—————————
CFMCategory=Baseline, one-step=true2026.053.1269.72————————
PAIRFLOWSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=82025.123.12—————————
UDLMSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=162025.123.12—————————
ARMEvaluator Model=Llama 3.2 3B2026.063.12————3.9430———
PAIRFLOWSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=162025.123.13—————————
UDLMSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=322025.123.13—————————
UDLMSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=642025.123.13—————————
UDLMSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=1282025.123.13—————————
UDLMSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=2562025.123.13—————————
UDLMSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=5122025.123.13—————————
UDLMSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=10242025.123.13—————————
PAIRFLOWSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=322025.123.14—————————
PAIRFLOWSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=642025.123.15—————————
PAIRFLOWSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=1282025.123.15—————————
PAIRFLOWSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=2562025.123.15—————————
PAIRFLOWSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=5122025.123.15—————————
PAIRFLOWSequence Length (N)=32, Training Iterations (Iter.)=200k, Number of Function Evaluations (NFE)=10242025.123.16—————————
UDLMSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=42025.123.57—————————
PAIRFLOWSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=42025.123.57—————————
Coupling ModelsCategory=Ours, Temperature=0.50, top-p=0.95, latent scale zscale=0.8, one-step=true2026.053.608845.82————————
UDLMSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=82025.123.63—————————
PAIRFLOWSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=82025.123.65—————————
UDLMSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=162025.123.67—————————
UDLMSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=322025.123.68—————————
PAIRFLOWSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=162025.123.69—————————
UDLMSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=642025.123.69—————————
UDLMSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=5122025.123.69—————————
PAIRFLOWSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=322025.123.7—————————
UDLMSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=1282025.123.7—————————
UDLMSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=2562025.123.7—————————
UDLMSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=10242025.123.7—————————
MDMEvaluator Model=Llama 3.2 3B2026.063.7————4.8185———
PAIRFLOWSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=642025.123.71—————————
PAIRFLOWSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=1282025.123.71—————————
PAIRFLOWSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=2562025.123.71—————————
PAIRFLOWSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=10242025.123.71—————————
PAIRFLOWSequence Length (N)=64, Training Iterations (Iter.)=400k, Number of Function Evaluations (NFE)=5122025.123.72—————————
Duo + Di4CCategory=Baseline, one-step=true2026.053.79292.94————————
DFM (ESD)Category=Baseline, one-step=true2026.053.7968.11————————
UDLMSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=42025.123.98—————————
PAIRFLOWSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=42025.124—————————
DFM (PSD)Category=Baseline, one-step=true2026.054.0694.08————————
Coupling ModelsCategory=Ours, Temperature=0.60, top-p=0.95, latent scale zscale=0.8, one-step=true2026.054.084176.19————————
UDLMSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=82025.124.09—————————
PAIRFLOWSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=82025.124.11—————————
UDLMSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=162025.124.14—————————
PAIRFLOWSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=162025.124.16—————————
UDLMSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=322025.124.16—————————
FMLMCategory=Baseline, one-step=true2026.054.16119.34————————
UDLMSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=642025.124.17—————————
UDLMSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=1282025.124.17—————————
UDLMSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=2562025.124.17—————————
PAIRFLOWSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=322025.124.18—————————
UDLMSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=5122025.124.18—————————
UDLMSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=10242025.124.18—————————
PAIRFLOWSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=642025.124.19—————————
PAIRFLOWSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=5122025.124.19—————————
PAIRFLOWSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=1282025.124.2—————————
PAIRFLOWSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=2562025.124.2—————————
PAIRFLOWSequence Length (N)=128, Training Iterations (Iter.)=600k, Number of Function Evaluations (NFE)=10242025.124.2—————————
MDLM + SDTTCategory=Baseline, one-step=true2026.054.311,429.48————————
LM1B test setCategory=Reference2026.054.3147.47————————
Coupling ModelsCategory=Ours, Temperature=0.65, top-p=0.95, latent scale zscale=0.8, one-step=true2026.054.319298.96————————
Duo + DCDCategory=Baseline, one-step=true2026.054.331,224.52————————
MDLM + Di4CCategory=Baseline, one-step=true2026.054.381,217.1————————
Coupling ModelsCategory=Ours, Temperature=0.70, top-p=0.95, latent scale zscale=0.8, one-step=true2026.054.5716134.06————————
ARBatch Size=42026.06—104.8——————0.541
ARBatch Size=162026.06—104.8——————2.031
ARBatch Size=1282026.06—104.8——————15.41
BERT-MouthPretrained=true2022.11—142.890.28670.124——————
BERT-MouthSteps=64, Inference Time (secs)=2.182022.11—142.89————————
BERT-MouthSteps=512, Inference Time (secs)=14.392022.11—86.78————————
D3PMPretrained=false, Schedule=(T - t + 1)^-1, Time Step=LTE2022.11—82.340.38970.2347——————
D3PMPretrained=false, Schedule=(T - t + 1)^-1, Time Step=TAD2022.11—125.150.3390.272——————
D3PMPretrained=false, Schedule=Spindle, Time Step=LTE2022.11—77.50.42410.2288——————
D3PM absorbinginference steps=1000, context window length=128, layers=12, vocabulary size=8192, number of seeds=22021.07—76.9——1.9—————