All issues

Search results

[Issue]: Gemma4 tool-call parsing produces wrong name and arguments (v1.0.5)

Status: Open.

#722 In ROCm/FastFlowLM;

· skarpov2 opened on Sep 10, 2026

Flm with gemma 4 has very poor ability to call tools

Status: Open.

#721 In ROCm/FastFlowLM;

· tidotua opened on Sep 10, 2026

Inconsistent memory consumption?

Status: Open.

#720 In ROCm/FastFlowLM;

· VladimirVLF opened on Sep 10, 2026

[Issue]: JSON literal data types instead of strings

Status: Open.

#719 In ROCm/FastFlowLM;

· gerritb opened on Sep 10, 2026

The directory for models is not configurable in the new MSI installer

Status: Open.

#718 In ROCm/FastFlowLM;

· antrv opened on Sep 10, 2026

Server silently serves the resident model for unresolvable model tags, echoing the requested name

Status: Open.

#716 In ROCm/FastFlowLM;

· Platano78 opened on Sep 9, 2026

[Feature]: Support MiniCPM5-2B top-ranked open 2B model

Status: Open.

#715 In ROCm/FastFlowLM;

· laurentpayot opened on Sep 8, 2026

Qwen3-8B on XDNA2 (Strix Halo): only ~10 tok/s decode + slow TTFT vs GPU - is this expected?

Status: Open.

#713 In ROCm/FastFlowLM;

· dg-w36 opened on Sep 8, 2026

[Model Support]: Unofficial MiniCPM5-2B Q4NX pack for XDNA 2

Status: Open.

#712 In ROCm/FastFlowLM;

· julianmb opened on Sep 8, 2026

[Feature]: Publish signed binaries

Status: Open.

#711 In ROCm/FastFlowLM;

· Daft-Wullie opened on Sep 7, 2026

Qwen3.6-35B-A3B Linux v1.0.4: ~2.7s fixed per-request TTFT = ~4,900 serialized per-expert NPU execs — validated coarsening fix (concatenated-expert GEMM)

Status: Open.

#710 In ROCm/FastFlowLM;

· bong-water-water-bong opened on Sep 6, 2026

[Linux] Weight bos are MAP_LOCKED — models >~18GB cannot load on 22.6GB RAM machines and some request

Status: Open.

#709 In ROCm/FastFlowLM;

· Cero-Xspr opened on Sep 5, 2026