AI Dataset2026
A multiple-choice benchmark for evaluating language-model arithmetic: 1,000 continuation-style elementary word problems (4 choices, balanced labels) organized by topic, grade band, and difficulty. Designed for base-model continuation log-likelihood scoring; released under Apache-2.0.
AI Dataset2026
A curated collection of 49,772 teacher-generated chat traces from qwen3.8-max-preview for supervised fine-tuning and off-policy distillation. Preserves visible chain-of-thought blocks, emphasizes math/code/reasoning mixes, and includes provenance and licensing cautions tied to Alibaba Cloud Model Studio.
Performs full-parameter post-training of trillion-parameter MoE DeepSeek-V4 models on an Ascend NPU SuperPOD, using a hierarchical optimization of model parallelism, communication orchestration, and kernel execution to increase Model FLOPs Utilization. Also builds CPT/SFT pipelines with solver-verified synthetic data for Operations Research, reporting strong zero-shot Pass@1 results.
AI Dataset2026
Provides a 57,937-row, quality-filtered multi-teacher SFT distillation corpus combining outputs from Qwen3.8-Max, GLM-5.2 and Kimi K3 across math, code, reasoning, tool-use and dialogue. Includes 24 parquet training views (including a pre-tokenized GLM-4.7 view), configurable sampling weights (sft_balanced), and explicit tool-call trajectories for agent training.
AI Dataset2026
Provides large-scale mathematical problem-solving, rewriting, and dialogue data organized into five Parquet-backed subsets for reasoning-oriented language-model training. Subsets support streaming access, Dataset Viewer inspection, and per-subset provenance metadata; licensed Apache 2.0.
Selectively admits dense token-level teacher supervision only after a prompt-level verifier audit, routing prompts that fail the audit to verifier-grounded trajectory supervision instead — reducing harmful updates from confidently wrong teachers and improving teacher GPU utilization.
AI Dataset2026
Provides a bilingual Chinese–English corpus for LLM training covering pretraining, capability-oriented midtraining (16K–256K long contexts), and supervised fine-tuning. Includes ~4.2T pretrain tokens, ~600B midtrain tokens, and ~4.57M SFT samples; sources span web, PDFs/OCR, code, math, QA, and agentic trajectories under mixed upstream licenses.
AI Dataset2026
Provides ~86K verifiable-reward RL training samples across Math, Knowledge (STEM), Long-Context, and Code for post-training LLMs; each sample includes a ground-truth and verifier-friendly JSONL format for stable reward signals. Note: code tasks require an external sandbox to execute tests.
Presents an open 7B dense foundation model for mathematical reasoning and tool-assisted web search, supporting 256K-token context. Key features: hybrid gated sliding-window + global attention for efficient long-context training, FP8 Muon optimizer, progressive 16K→64K→256K curriculum, and open-source weights, data, and recipes.
Analyzes why on-policy distillation can produce excessively long student rollouts by identifying termination-token mismatch between students and teachers, and proposes aggregating functionally equivalent EOS tokens as a shared semantic stopping action to mitigate length inflation.