3+ hour, 15+ min ago   (263+ words) This article opens up that layer. Compaction, memory strategy, context budgeting, and todo-state are the machinery that turns a shallow loop into a deep agent. We look at how LangChain Deep Agents, Claude Code, Manus, OpenAI Codex, and Amazon Bedrock…...

2+ week, 3+ day ago   (155+ words) Yes, on two tracks. The weights are live on Hugging Face under an MIT license, and a hosted API is already priced and serving. GLM-5.3-Flash starts from a newly trained base model on a 30T-token multimodal corpus. Three changes…...

2+ week, 3+ day ago   (336+ words) Yes but not on a workstation. The FP8 checkpoint is 172.78 GiB and the BF16 checkpoint is 335.28 GiB. Per vLLM recipes, TP2 is the minimum validated FP8 configuration on GB300 and TP4 is recommended. On an 8×H200 node, use TEP8; plain TP8 is incompatible with the checkpoint’s 128-wide quantization blocks....

2+ week, 4+ day ago   (303+ words) Yes, All three Granite 4.2 language models ship under Apache 2.0, so download, fine-tuning, and commercial production use carry no licensing gate. Granite 4.2 is a decoder-only dense transformer, not a hybrid or MoE design. Core components are Grouped Query Attention with 8 KV…...

3+ week, 19+ hour ago   (565+ words) Most teams treat ‘which model’ as the important decision. The harness engineering literature keeps pointing somewhere else. In LangChain’s Terminal-Bench experiment, changing only the harness—same model throughout—moved a coding agent from roughly 30th place into the top 5. The center…...

3+ week, 2+ day ago   (398+ words) Yes, if you self-host. The weights ship as Safetensors and GGUF, and the drafter checkpoints are not served by any hosted inference provider on Hugging Face today. Running them needs an SGLang or llama.cpp build with DSpark support for…...

4+ week, 21+ hour ago   (574+ words) Implement an end-to-end fine-tuning pipeline for tool-calling language models. This tutorial covers parsing trajectories, structured tool-call extraction, Qwen-compatible ChatML rendering, and efficient LoRA adaptation using PyTorch. The post Fine-Tuning Tool-Calling LLMs: A Complete Guide Using XYZ-Aquila-SFT and Qwen3 appeared first on…...

4+ week, 2+ day ago   (212+ words) Partially, GLM-5.3 is live through the Z.ai API, the GLM Coding Plan, and ZCode. Weights are not out. Z.ai says it will publish them roughly two weeks after launch, once safety evaluation and hardening finish. Terminal-Bench 3.0 moves from 4.6 to 28.3 against…...

4+ week, 2+ day ago   (393+ words) Yes, Needle 2 ships as prebuilt binaries and a static library for macOS, Linux (x86-64, ARM64, ARMv7, RISC-V, MIPS32el), Windows, Android, iOS/watchOS/tvOS, and WebAssembly. Cactus says Pebble already runs Needle locally in the Index 01 app for offline voice actions. Needle 2 uses what Cactus…...

1+ mon, 2+ day ago   (252+ words) Four stages sit on top of the base model. LoRA supervised fine-tuning on a synthetic formal-logic corpus. Checkpoint fusion, averaging intermediate SFT checkpoints in parameter space. WiSE-FT interpolation back toward the pretrained base at λ = 0.25. Then MGPO, an entropy-weighted GRPO…...