All issues

Search results

[Bug] A single padding token can trigger Math SDPA fallback and a quadratic GPU-memory spike [flash_attn_varlen_func not supported]

Status: Open.

#136 In ML-GSAI/LLaDA;

· zpointS opened on Aug 9, 2026

Verify evals on Papers with Code

Status: Open.

#135 In ML-GSAI/LLaDA;

· NielsRogge opened on Jul 22, 2026

Segfault loading LLaDA-8B-Instruct under transformers >= 5 (new core_model_loading path) — works with <5 pin

Status: Open.

#133 In ML-GSAI/LLaDA;

· ghost opened on Jul 10, 2026

LLaDA inference logits differ across NVIDIA GPUs (Blackwell vs 4090) with same weights/code

Status: Open.

#131 In ML-GSAI/LLaDA;

· Kevin-lkw opened on Apr 13, 2026

Typo in Algorithm 5: Masking ratio inverted?

Status: Open.

#129 In ML-GSAI/LLaDA;

· lazitech opened on Mar 15, 2026

📋 Documentation Enhancement Suggestion

Status: Open.

#128 In ML-GSAI/LLaDA;

· croviatrust opened on Mar 10, 2026

Usage of CFG for Conditional Generation Benchmarks

Status: Open.

#125 In ML-GSAI/LLaDA;

· JeiminJeon opened on Feb 19, 2026

[Docs] Add PRISM test-time scaling recipe / reference for LLaDA-8B-Instruct in README

Status: Open.

#123 In ML-GSAI/LLaDA;

· viiika opened on Feb 4, 2026

How to evaluate the peft model

Status: Open.

#122 In ML-GSAI/LLaDA;

· zaiquanyang opened on Jan 2, 2026

Unconditional Generation

Status: Open.

#121 In ML-GSAI/LLaDA;

· Yuxuan314 opened on Dec 24, 2025

RoPE and attention mask mismatch

Status: Open.

#120 In ML-GSAI/LLaDA;

· probablyabot opened on Dec 23, 2025

cpu memory OOM when using lm-eval to evaluate LLADA

Status: Open.

#119 In ML-GSAI/LLaDA;

· duterscmy opened on Dec 4, 2025