Search results

implement this spinfoam environment

Status: Open.

#429 In GFNOrg/torchgfn;

· josephdviviano opened on Nov 3, 2025

Extend modules to support more kinds of diffusion processes

Status: Open.

#422 In GFNOrg/torchgfn;

· josephdviviano opened on Oct 23, 2025

Possible variants of the DiffusionSampling environment to add

Status: Open.

#421 In GFNOrg/torchgfn;

· josephdviviano opened on Oct 23, 2025

Add policy gradient method

enhancementNew feature or requestNew feature or request

high priorityLet's do these first!Let's do these first!

Status: Open.

#416 In GFNOrg/torchgfn;

· josephdviviano opened on Oct 14, 2025

Explore puttingget_trajectory/transition_* back into the estimator via a PolicyMixin

Status: Open.

#414 In GFNOrg/torchgfn;

· josephdviviano opened on Oct 14, 2025

Some sources of inefficiency.

Status: Open.

#377 In GFNOrg/torchgfn;

· hyeok9855 opened on Aug 19, 2025

Thompson sampling for improved exploration in GFlowNets

enhancementNew feature or requestNew feature or request

Status: Open.

#374 In GFNOrg/torchgfn;

· josephdviviano opened on Aug 10, 2025

Improving Gradient-guided Nested Sampling for Posterior Inference

enhancementNew feature or requestNew feature or request

Status: Open.

#373 In GFNOrg/torchgfn;

· josephdviviano opened on Aug 10, 2025

pass off policy hyperparameters as a vector with the length of a batch instead of a scalar

Status: Open.

#369 In GFNOrg/torchgfn;

· josephdviviano opened on Aug 1, 2025

Experiment with different precision levels for performance?

Status: Open.

#366 In GFNOrg/torchgfn;

· josephdviviano opened on Jul 30, 2025

Need a way of calculating posterior recovery in very high dimensional space -- i.e., by pre-computing all states once and saving these results to disk.

Status: Open.

#365 In GFNOrg/torchgfn;

· josephdviviano opened on Jul 30, 2025

Method for sampling different off-policy exploration techniques (i.e., strategies for each policy to adopt)

Status: Open.

#361 In GFNOrg/torchgfn;

· josephdviviano opened on Jul 30, 2025