Recurrent Models - a tomg-group-umd Collection

tomg-group-umd 's Collections

Refusal Token Models

Efficient_inference_models

Gemstone Models

Recurrent Models

Style Descriptors

GenQA

Goldfish Loss: Mitigating Memorization in LLMs

Zero-Shot Grafting

Recurrent Models

updated May 21

These are checkpoints for recurrent LLMs developed to scale test-time compute by recurring in latent space.

tomg-group-umd/huginn-0125

Text Generation • 4B • Updated Jul 29 • 3.33k • 283
Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach

Paper • 2502.05171 • Published Feb 7 • 150
tomg-group-umd/huginn_swa_100_10_avg_0.9_merge

Text Generation • 4B • Updated Jul 17 • 3
tomg-group-umd/step-00010752-recurrence_full_512_0

Text Generation • 4B • Updated Jul 17 • 3
tomg-group-umd/step-00041728-recurrence_full_512_0

Text Generation • 4B • Updated Jul 17 • 3
tomg-group-umd/step-00006144-baseline_2_0

Text Generation • 4B • Updated Jul 17 • 3
tomg-group-umd/step-00010720-baseline_2_0

Text Generation • 4B • Updated 17 days ago • 12
tomg-group-umd/step-00035840-recurrence_full_512_0

Text Generation • 4B • Updated Jul 17 • 3
tomg-group-umd/step-00029824-recurrence_full_512_0

Text Generation • 4B • Updated Jul 17 • 3
tomg-group-umd/step-00023808-recurrence_full_512_0

Text Generation • 4B • Updated Jul 17 • 3
tomg-group-umd/step-00017920-recurrence_full_512_0

Text Generation • 4B • Updated Jul 17 • 3
tomg-group-umd/step-00011904-recurrence_full_512_0

Text Generation • 4B • Updated Jul 17 • 15
tomg-group-umd/step-00006144-recurrence_full_512_0

Text Generation • 4B • Updated Jul 17 • 4
tomg-group-umd/huginn_swa_75_7_ema_0.9_merge

Text Generation • 4B • Updated Jul 17 • 963 • 1
tomg-group-umd/huginn-dataset

Viewer • Updated Jul 15 • 274M • 2.88k • 6