Search results

Unguarded allocation sites in candle-core/src/quantized/gguf_file.rs (DoS via crafted GGUF)

Status: Open.

#3533 In huggingface/candle;

· UNILESS opened on May 14, 2026

Candle leaves ~ model size of memory unused on Spark

Status: Open.

#3530 In huggingface/candle;

· sempervictus opened on May 11, 2026

Cannot load tokenizer from gguf file

Status: Open.

#3527 In huggingface/candle;

· wangyang1749 opened on May 9, 2026

Cross-CUDA Tensor::to_device fails with CUDA_ERROR_INVALID_CONTEXT (no peer-access enable)

Status: Open.

#3524 In huggingface/candle;

· toddwbucy opened on May 8, 2026

Phi4 mini multiple runs

Status: Open.

#3523 In huggingface/candle;

· dimitrov-anasoft opened on May 8, 2026

Metal SDPA panics during quantized LLaMA GQA decode

Status: Open.

#3522 In huggingface/candle;

· athola opened on May 7, 2026

candle-flash-attn vendored kernels ~13 months stale; long-context bf16 divergence vs PyTorch flash_attn 2.8.3

Status: Open.

#3515 In huggingface/candle;

· toddwbucy opened on May 6, 2026

Contribute Qwen3.5/3.6 35B-A3B port + 3 fused perf kernels — packaging guidance?

Status: Open.

#3514 In huggingface/candle;

· lukeme117 opened on May 6, 2026

Qwen3 VL forward pass doesn't construct attention mask properly

Status: Open.

#3505 In huggingface/candle;

· raka-gunarto opened on May 1, 2026

The fact that Candle supports Metal is not mentioned in the mdbook (installation)

Status: Open.

#3497 In huggingface/candle;

· Xuanmizhen opened on Apr 27, 2026

Gemma 4 Textmodel doesn't work.

Status: Open.

#3492 In huggingface/candle;

· AlpineVibrations opened on Apr 25, 2026

[candle-onnx] simple_eval is hardcoded to CPU, preventing CUDA/GPU acceleration

Status: Open.

#3491 In huggingface/candle;

· UU-0665D27 opened on Apr 23, 2026