硬體

AI 晶片、GPU、資料中心、邊緣裝置與硬體安全。

Ollama Community Proposes Streaming Order Fix After Text-to-Tool Calls May Be Rejected by Responses Clients

A community report on October 5 says Ollama may reuse output indexes and omit a message completion event when it emits text before a tool call. The proposed fix includes regression tests but has not been merged or validated with actual model inference.

2026-10-03

GitHub Repo

Transformers Community Reports FSDP2 Loading Memory Bug; Multi-GPU Startup May Create Duplicate Full Models

A community report involving Transformers 5.12.0 says that after enabling the CPU memory-efficient loading option, other training processes may still allocate a full model, exhausting host memory. Official documentation says the option should prevent duplicate loading across processes, but the root cause and fix status await upstream confirmation.

2026-10-05

GitHub Repo

PyTorch Community Reports CUDA Interpolation Shortcut Bug That May Skip Resampling for Same-Size Outputs

A reproduction on a PyTorch development build shows that CUDA bicubic interpolation may return the input unchanged when a non-integer scale factor still produces the same output dimensions. A fix has been proposed but not merged; the impact on released versions remains unconfirmed.

2026-10-04

GitHub Repo

vLLM Community Reports Batch Invariance Cold-Start Bug That May Consistently Produce Incorrect Tokens on Specific Inference Paths

A reproduction on the vLLM development branch shows that, even with batch invariance enabled, model metadata caching and module import order may still change inference results. The community has proposed a fix, but it has not yet been merged; the impact on released versions and other hardware remains to be confirmed.

2026-09-27

GitHub Repo

Ollama Community Reports Missing Text in LFM2 Output; Certain Python Strings May Disappear Silently

A community reproduction on Ollama 0.35.1 suggests that `lfm2:24b` may omit certain Python strings, altering JSON values and document content. Upstream has yet to confirm the issue, and there is not enough evidence to determine whether it stems from tokenizer decoding, model packaging, or the hardware path.

2026-10-04

GitHub Repo

vLLM Community Reports Silent Loss of Image Input; Per-Request Thinking Settings May Affect Visual Understanding

A community report involving vLLM 0.30.0 suggests that adding `chat_template_kwargs` to image requests may leave the model with text only, while the server still returns success. The issue has yet to be confirmed upstream; deployments combining visual input with thinking-mode controls should verify the actual input.

2026-10-04

GitHub Repo

Ai2 Open-Sources Olmo-core 3, Redesigning Expert Placement and Data Routing for MoE Training

The new version keeps experts resident on GPUs and uses data routing to reduce the cost of repeatedly aggregating weights. Initial tests reported about 2.7× throughput. A trillion-parameter configuration has passed systems testing, but it used random routing, so training quality and stability remain unverified.

2026-10-03

GitHub Repo

vLLM Community Reports Output Divergence After Restart; Core Autotuning May Affect Reproducibility

Tests with Qwen3-8B, a fixed seed, and zero temperature still produced different outputs when core configurations varied. A related PyTorch patch has not been merged, and the scope of the issue needs to be checked against the version, hardware, and compilation path.

2026-09-27

GitHub Repo

vLLM Community Test: Enabling Pinned Memory in WSL2 Cuts Single-GPU Decoding Latency by About 12%

A single-machine test suggests that changing a memory setting can reduce GPU wait time during decoding. The default remains disabled; benefits and capacity limits should be validated in each deployment environment.

2026-09-27

GitHub Repo

vLLM Community Reports Structured Output Bug That Can Rewrite JSON Constants Under Certain Settings

Under certain settings, the Guidance backend rewrites constant objects, causing valid outputs to be rejected and invalid variants to pass. The community has submitted a fix and tests, but the change has not been merged, and the affected version range remains unclear.

2026-09-26

AI 研究

AlphaFold Database Adds 8,028 Predicted Viral Protein Complexes as Inference Pipeline Goes Open Source

The collaboration released high-confidence complex predictions for 2,812 viral proteomes, along with a GPU computing pipeline. Deployment still requires large GPUs and cluster infrastructure, and predicted structures need experimental validation.

2026-09-26

GitHub Repo

ServingStudio Introduces an Open Inference Performance Workbench for Simulation, Agent-Driven Changes, and GPU Validation

The system uses measured GPU kernel times to predict serving bottlenecks, then has an agent implement candidate optimizations in an inference framework. The team has released the code and test cases, but gains depend on the model and workload, and some patches are still awaiting upstream review.

2026-09-26