Tag

MiniCPM5-2B

Compact causal LLM for on-device assistants, coding agents and long-context tool use — ~2.52B parameters with a 131,072-token context, trained with SFT + RL + OPD and released with its UltraData training corpora and multi-format deployment checkpoints.

CodeMidas: Scaling Agentic Coding RL Environments from Code Itself

Turns implemented functionality in open-source codebases into executable RL training environments by automatically generating behavioral specifications, execution-grounded tests, and post-rollout filtering; yields 5,545 verifiable coding tasks across 23 languages to train coding agents.

MiMo-V2.6-Flash-RL

A sparse-MoE multimodal LLM fine-tuned with large-scale mixed reinforcement learning, supporting text/image/video/audio and a 1M-token context window; designed for long-horizon agentic tasks, coding and visual reasoning with groupwise grading and on-policy distillation.

MiMo-V2.6-Distill-Qwen-9B

A 9B agentic multimodal SFT checkpoint distilled from Qwen3.5-9B for coding, general agent tasks, visual coding and cybersecurity. Provided by Xiaomi MiMo as a research seed (77.4B-token SFT mix) to bootstrap agentic RL and tool-use experiments.

Post-Training Leaves Behavioral Shadows on Unrelated Decisions

Shows that private post-training changes leave measurable "behavioral shadows" in task-unrelated single-word outputs and proposes Active Taskless Distillation (ATD) to transfer capabilities to a student using only those single-word teacher responses, without teacher logits or parameters.

OrcaSAQ2 27B

Compresses Qwen3.8-27B into a 12.3 GB sensitivity-aware mixed-precision quantized checkpoint for long-horizon agent workloads; preserves BF16 fidelity (+0.02% PPL, 93.2% token Top‑1 agreement), supports 262K context and vLLM serving, text-only and Apache‑2.0 licensed.

XiaomiMiMo/MiMo-V2.6-RL-oss

Provides mixed-domain, verifiable RL training environments for LLM agents (code, cyber, knowledge work, web dev, music) as Parquet datasets, with domain-specific verifiers, Docker artifacts and links to training code for reproducible agentic RL experiments.

Qwen3.8-Flash-Next · GSQ-RCO Coder

Capability-targeted compression of Qwen3.8-Flash-Next: half the experts are removed and remaining weights quantized to 3.5 bpw, producing a 58.4 GB GGUF (29.6 GB resident) that preserves coding and multimodal ability while trading off other domains.

Naive-N0.5-Flash

Open-weight 309B Mixture-of-Experts causal LLM with 15.5B active parameters and a native 1M-token context for coding and AI R&D. Combines Sliding-Window Attention and DeepSeek Sparse Attention (no full-attention layers), supports FP8 inference; weights under MIT license.

Kolibri-1

Processes English and German text with long-context reasoning and structured tool-calling. Uses a 78B mixture-of-experts architecture that activates ~3.46B parameters per token, offers native 262k-token context (validated to 1M), and is released as Apache-2.0 weights — suited for RAG, document processing and human-in-the-loop decision support.

FreeInference Agentic Trace

Provides 16 weeks of anonymized production agent-session traces (12,002 sessions, ~1.19M LLM requests, ~1.21M tool calls, 209B input tokens) released as block-level prefix IDs plus flattened Parquet tables for KV-cache, scheduling and serving-system research.

TRACE: Rollout-Guided Quantization-Aware Training for FP4 Reinforcement Learning of MoE Language Models

Proposes TRACE, an FP4 quantization framework for RL of MoE LLMs that uses rollout-side FP4 outcomes to guide training-side rounding and caches mantissa/scale from deeper layers to limit overhead. Preserves BF16-level RL performance while enabling up to 5.4× rollout speedup.