All issues

Search results

qwen2 (DeepSeek-R1-Distill) produces gibberish on iOS CPU path; same llama.cpp generation clean on macOS CPU

Status: Open.

#374 In mybigday/llama.rn;

· KA-devl opened on Jul 30, 2026

Vulkan support for Android

Status: Open.

#364 In mybigday/llama.rn;

· roshansoni-io opened on Jul 11, 2026

[Feature Request] Support draftless n-gram speculative decoding (ngram-mod / --spec-default)

Status: Open.

#359 In mybigday/llama.rn;

· abhinavk454 opened on Jun 28, 2026

mutlimodel issues

Status: Open.

#356 In mybigday/llama.rn;

· animesh65432 opened on Jun 18, 2026

The inference fail on Qualcomm NPU

Status: Open.

#353 In mybigday/llama.rn;

· ZhangWei125521 opened on May 28, 2026

SIGSEGV in lm_ggml_compute_forward_mul_mat+22XX → __memcpy_aarch64_simd on aarch64 / Samsung Exynos under sustained completion load (Qwen3-0.6B-Q4_K_M, llama.rn 0.12.3 / llama.cpp b9254) — refines #336

Status: Open.

#350 In mybigday/llama.rn;

· eckonode opened on May 24, 2026

GLM-OCR fails to load — "Failed to load model" on all configurations

Status: Open.

#337 In mybigday/llama.rn;

· tutak opened on May 6, 2026

First ctx.completion() throws Exception in HostFunction: <unknown> on Samsung Exynos (qwen3-0.6b / TinyLlama / gemma4-e2b-lite, both with and without no_extra_bufts: true)

Status: Open.

#336 In mybigday/llama.rn;

· eckonode opened on May 2, 2026

Add QNN backend support

Status: Open.

#334 In mybigday/llama.rn;

· BlindDeveloper opened on Apr 27, 2026

Suppport for TurboQuant

Status: Open.

#326 In mybigday/llama.rn;

· youngrichu opened on Apr 9, 2026

Qwen 3.5 models generates only zeroes on A14 chips

Status: Open.

#305 In mybigday/llama.rn;

· navedmerchant opened on Mar 10, 2026

Not Work on SOC SM7450 Snapdragon 7 Gen 1

Status: Open.

#279 In mybigday/llama.rn;

· matrixportalx opened on Jan 19, 2026