#开源模型

共 111 篇带此标签

111 articles Newest 2026-11-21
Archive · #开源模型 11 articles
LLMs

Tencent Hy4 Flash: More Than Just Speed

Hy4 Flash reached the Hacker News front page with 72 points and 11 comments. The real questions for developers are whether it can complete Agent tasks reliably at lower cost, how it can be deployed, and whether its weights are openly licensed.

2026-11-18 Source 2026-11-17 5 min 58 views
LLMs

Can Tencent’s Rebuilt Hy4 Architecture Be Reproduced?

Hy4 reached the Hacker News front page, but its official summary omits model parameters, training details, code, weights, and benchmark data. This analysis outlines the reproducibility checklist a model release should meet and what Tencent still needs to disclose.

2026-11-16 Source 2026-11-04 7 min 58 views
Opinion

Can Model Weights Carry Viruses? LLM Supply Chain Risks

Malicious patterns may be embedded in model weights and propagate through downloads, quantization, fine-tuning, and redistribution. Although the current evidence remains limited, production teams should pin model sources, record hashes, and audit the exact weights they deploy.

2026-11-15 Source 2026-11-13 7 min 58 views
LLMs

Choosing Qwen3.8-27B Quantization by VRAM

A practical framework for comparing Qwen3.8-27B quantizations across VRAM limits, using Unsloth's Dynamic v3.1 guidance and reproducible benchmarks for speed, memory use, output quality, and framework compatibility.

2026-11-15 Source 2026-11-14 8 min 58 views
Case Studies

GLM-5.3 on Two DGX Stations: 607.8 tok/s

NVIDIA deployed GLM-5.3 across two DGX Station systems with vLLM, reporting 607.8 tok/s throughput. The setup shows how multi-node inference expands capacity while leaving important production benchmarks unanswered.

2026-11-15 Source 2026-11-14 6 min 58 views
Tools

How Maple Fits Gemma 4 Into 6GB of VRAM

Maple applies Sub-2-BPW Moe Pruning to Gemma 4 26B, with claims of a 6GB GPU footprint and 45 tokens per second. The headline figures are intriguing, but independent replication and pruning-quality data remain limited.

2026-11-08 Source 2026-11-08 6 min 58 views
Opinion

How to Choose an AI Agent Model by Cost per Task

Village.dev argues that closed models now lead on price-performance. For Agent teams, however, the relevant metric is the total cost of each successfully completed task, including retries, tool calls, and human intervention.

2026-11-06 Source 2026-11-06 8 min 58 views
LLMs

ERNIE 5.5 Flash Base: Open Weights, Hidden Costs

Open weights do not automatically make inference cheaper. This analysis examines ERNIE 5.5 Flash Base through tool use, tail latency, GPU utilization, and context management to assess whether it merits enterprise deployment.

2026-11-06 Source 2026-11-05 9 min 58 views
Case Studies

What $20,000 Really Buys for an Open-Weight Image Model

A two-person lab trained the PJIT-21K text-to-image model from scratch for roughly $20,000 and released its weights under Apache 2.0. This analysis breaks down training, inference, licensing, and deployment costs for teams evaluating open models.

2026-11-06 Source 2026-11-04 7 min 58 views
LLMs

Tencent's Hy4 Open-Source Launch Tests Its Ecosystem

Hy4 reached the Hacker News front page with 68 points and one comment, giving Tencent early visibility among developers. Whether it becomes a leading open-source model will depend on its license, inference costs, tooling, and real-world adoption.

2026-11-06 Source 2026-11-05 6 min 58 views