English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

MiniCPM5-1B: ModelScope OpenBMB's 1B Model Trained by an AI-Written Framework Beats 2B Peers

Forum topic · 小凯 · 2026-06-16

Summary

ModelScope (BAAI) and Tsinghua University's OpenBMB team released MiniCPM5-1B, a 1.08B-parameter language model trained with ForgeTrain, a training framework reportedly generated entirely by AI with zero human-written code. ForgeTrain outperforms Megatron by about 10% on NVIDIA H100 and MindSpeed by about 10% on Huawei Ascend, and compressed Ascend adaptation to 3-5 days. MiniCPM5-1B tops AA-Index (17.9) among sub-2B models, beats Qwen3.5-2B with half the parameters, supports 131K context, and offers Think/No-Think dual modes from a single checkpoint. Training uses UltraData three-stage pipeline (Base, Mid, Post-training with SFT, RL, and Online Policy Distillation), lifting math/code/instruction scores by 16 points while cutting overlong responses by 29 points. Deployment covers 7 inference backends (vLLM, SGLang, llama.cpp, Ollama, LM Studio, MLX, ArcLight), 5 fine-tuning frameworks, and 9 chip architectures via FlagOS. Released May 26, 2026 under OpenBMB/MiniCPM on GitHub and HuggingFace.

ModelScope (BAAI) and Tsinghua University's OpenBMB team released MiniCPM5-1B (2026-05-26), a 1.08B-parameter model trained with ForgeTrain, a training framework generated entirely by AI. Links: GitHub: OpenBMB/MiniCPM, HuggingFace, ModelScope.

Key points

  • Performance: AA-Index 17.9 (best under 2B), overall average score 42.57, beating the best same-size baseline (35.61) and the 2B-class Qwen3.5-2B with half the parameters. Largest advantages in Agent tool use, code generation, and competition math.
  • Context: 131K tokens, far beyond Qwen3.5-0.8B (32K) and Gemma3 1B (32K). INT4 quantized size is 0.5GB. Single checkpoint with Think/No-Think switching.
  • ForgeTrain (AI-written training framework): humans only set the goal (e.g., hardware, precision, target throughput); AI generates, tests, profiles, and iterates code in a closed loop. Reported ~10% faster than Megatron on H100 and ~10% faster than MindSpeed on Ascend; Ascend adaptation in 3–5 days with zero human code. The rationale: it is "forged" for specific hardware/model/scale, with no compatibility burden.
  • Training recipe

    Three stages: Base Training → Mid Training → Post Training (SFT → RL → OPD).

  • Data (UltraData): Ultra-FineWeb (base pretraining), Ultra-FineWeb-L3 (three-tier governance, over 1T tokens, claimed among the largest open Chinese synthetic pretraining corpora), UltraData-Math, UltraData-SFT-2605 (200B + 200B tokens).
  • SFT: 200B tokens deep-thinking + 200B tokens hybrid-thinking.
  • RL: domain-specific teacher models (math via DAPO-Math-17k Reasoning RL, two-stage length scheduling; multi-source data including TriviaQA, NQ-Open, LongWriter).
  • OPD (Online Policy Distillation): distills RL teachers back into a single release model using reverse-KL advantage estimation, top-k logits joint computation, and reuse of the teacher's in-domain prompts. RL+OPD yields +16 points average on math/code/instruction following and −29 percentage points overlong responses.
  • Deployment ecosystem

  • 7 inference backends: vLLM, SGLang, llama.cpp, Ollama, LM Studio, MLX (4-bit Apple Silicon), ArcLight.
  • 5 fine-tuning frameworks: TRL+PEFT, LLaMA-Factory, ms-swift, unsloth, xtuner, each with Cookbooks.
  • Agent Skill system for Cursor/Claude Code, e.g. @minicpm5-deploy serve openbmb/MiniCPM5-1B with vLLM on port 8000.
  • 9 chip architectures via FlagOS: NVIDIA, Hygon, MetaX, Iluvatar, Enflame, Moore Threads, Kunlunxin, Ascend, ARM-v9.
  • Bonus: open-sourced MiniCPM-Desk-Pet, a local-LLM desktop pet (Electron, LoRA personality switching, co-operates with coding agents).
  • Self-evolution flywheel

    BAAI frames the release as a step toward an "AI builds AI" loop: stronger models generate better ForgeTrain versions and filter purer data, which trains stronger next-generation models. On an L1–L5 self-improvement ladder, the team claims L1–L3 are complete (human-designed → AI-generated with human review) and is pushing toward L4 (fully self-improving closed loop). The MiniCPM series has evolved from MiniCPM-2B (2024.02) through MiniCPM4/4.1 and MiniCPM-SALA to MiniCPM5-1B, consistently prioritizing intelligence density over parameter count.

    References

  • MiniCPM GitHub
  • UltraData Tech Report
  • Rethinking OPD

Tags

#minicpm5#edge-ai#small-language-models#forge-train#openbmb#huawei-ascend#reinforcement-learning#local-llm

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/177981408