Posts by Collection

daily

Daily AI Digest — 2026-05-11

Published: May 10, 2026

Daily AI Digest — 2026-05-12

Published: May 11, 2026

Daily AI Digest — 2026-05-13

Published: May 12, 2026

Daily AI Digest — 2026-05-14

Published: May 13, 2026

Daily AI Digest — 2026-05-15

Published: May 14, 2026

Daily AI Digest — 2026-05-16

Published: May 15, 2026

Daily AI Digest — 2026-05-17

Published: May 16, 2026

Daily AI Digest — 2026-05-18

Published: May 17, 2026

Daily AI Digest — 2026-05-19

Published: May 18, 2026

Daily AI Digest — 2026-05-20

Published: May 19, 2026

Daily AI Digest — 2026-05-21

Published: May 20, 2026

Daily AI Digest — 2026-05-22

Published: May 21, 2026

Daily AI Digest — 2026-05-23

Published: May 22, 2026

Daily AI Digest — 2026-05-24

Published: May 23, 2026

Daily AI Digest — 2026-05-25

Published: May 24, 2026

Daily AI Digest — 2026-05-26

Published: May 25, 2026

Daily AI Digest — 2026-05-27

Published: May 26, 2026

Daily AI Digest — 2026-05-28

Published: May 27, 2026

Daily AI Digest — 2026-05-29

Published: May 28, 2026

Daily AI Digest — 2026-05-30

Published: May 29, 2026

Daily AI Digest — 2026-05-31

Published: May 30, 2026

Daily AI Digest — 2026-06-01

Published: May 31, 2026

Daily AI Digest — 2026-06-02

Published: June 01, 2026

Daily AI Digest — 2026-06-03

Published: June 02, 2026

Daily AI Digest — 2026-06-04

Published: June 03, 2026

Daily AI Digest — 2026-06-05

Published: June 04, 2026

Daily AI Digest — 2026-06-06

Published: June 05, 2026

Daily AI Digest — 2026-06-07

Published: June 06, 2026

Daily AI Digest — 2026-06-08

Published: June 07, 2026

Daily AI Digest — 2026-06-09

Published: June 08, 2026

Daily AI Digest — 2026-06-10

Published: June 09, 2026

Daily AI Digest — 2026-06-11

Published: June 10, 2026

Daily AI Digest — 2026-06-12

Published: June 11, 2026

Daily AI Digest — 2026-06-13

Published: June 12, 2026

Daily AI Digest — 2026-06-14

Published: June 13, 2026

Daily AI Digest — 2026-06-15

Published: June 14, 2026

Daily AI Digest — 2026-06-16

Published: June 15, 2026

Daily AI Digest — 2026-06-17

Published: June 16, 2026

Daily AI Digest — 2026-06-18

Published: June 17, 2026

Daily AI Digest — 2026-06-19

Published: June 18, 2026

Daily AI Digest — 2026-06-20

Published: June 19, 2026

Daily AI Digest — 2026-06-21

Published: June 20, 2026

Daily AI Digest — 2026-06-22

Published: June 21, 2026

Daily AI Digest — 2026-06-23

Published: June 22, 2026

Daily AI Digest — 2026-06-24

Published: June 23, 2026

Daily AI Digest — 2026-06-25

Published: June 24, 2026

Daily AI Digest — 2026-06-26

Published: June 25, 2026

Daily AI Digest — 2026-06-27

Published: June 26, 2026

Daily AI Digest — 2026-06-28

Published: June 27, 2026

Daily AI Digest — 2026-06-29

Published: June 28, 2026

Daily AI Digest — 2026-06-30

Published: June 29, 2026

Daily AI Digest — 2026-07-01

Published: June 30, 2026

Daily AI Digest — 2026-07-02

Published: July 01, 2026

Daily AI Digest — 2026-07-03

Published: July 02, 2026

Daily AI Digest — 2026-07-04

Published: July 03, 2026

Daily AI Digest — 2026-07-05

Published: July 04, 2026

Daily AI Digest — 2026-07-06

Published: July 05, 2026

Daily AI Digest — 2026-07-07

Published: July 06, 2026

Daily AI Digest — 2026-07-08

Published: July 07, 2026

Daily AI Digest — 2026-07-09

Published: July 08, 2026

Daily AI Digest — 2026-07-10

Published: July 09, 2026

Daily AI Digest — 2026-07-11

Published: July 10, 2026

Daily AI Digest — 2026-07-12

Published: July 11, 2026

Daily AI Digest — 2026-07-13

Published: July 12, 2026

Daily AI Digest — 2026-07-14

Published: July 13, 2026

Daily AI Digest — 2026-07-15

Published: July 14, 2026

Daily AI Digest — 2026-07-16

Published: July 15, 2026

Daily AI Digest — 2026-07-17

Published: July 16, 2026

Daily AI Digest — 2026-07-18

Published: July 17, 2026

Daily AI Digest — 2026-07-19

Published: July 18, 2026

Daily AI Digest — 2026-07-20

Published: July 19, 2026

Daily AI Digest — 2026-07-21

Published: July 20, 2026

Daily AI Digest — 2026-07-22

Published: July 21, 2026

Daily AI Digest — 2026-07-23

Published: July 22, 2026

Daily AI Digest — 2026-07-24

Published: July 23, 2026

Daily AI Digest — 2026-07-25

Published: July 24, 2026

Daily AI Digest — 2026-07-26

Published: July 25, 2026

Daily AI Digest — 2026-07-27

Published: July 26, 2026

Daily AI Digest — 2026-07-28

Published: July 27, 2026

Daily AI Digest — 2026-07-29

Published: July 28, 2026

Daily AI Digest — 2026-07-30

Published: July 29, 2026

Daily AI Digest — 2026-07-31

Published: July 30, 2026

Daily AI Digest — 2026-08-02

Published: August 01, 2026

Daily AI Digest — 2026-08-03

Published: August 02, 2026

Daily AI Digest — 2026-08-04

Published: August 03, 2026

Daily AI Digest — 2026-08-05

Published: August 04, 2026

Daily AI Digest — 2026-08-06

Published: August 05, 2026

Daily AI Digest — 2026-08-07

Published: August 06, 2026

Daily AI Digest — 2026-08-08

Published: August 07, 2026

llm-libs

PyTorch 深度学习框架

张量操作、autograd、nn.Module、优化器、分布式训练(DDP/FSDP)、混合精度(AMP)、CUDA

NumPy 科学计算库

ndarray、广播机制、矩阵运算、线性代数(SVD/特征值)、FFT、在LLM中的数据预处理

Pandas 数据处理库

DataFrame、数据读写、清洗、groupby聚合、merge/join、训练数据预处理

Pydantic 数据验证库

BaseModel、数据验证、序列化、Field约束、与LLM结构化输出配合

tiktoken 分词库

BPE分词、encoding_for_model、token计数与成本估算、cl100k_base/p50k_base编码

tokenizers 分词库

BPE/WordPiece/Unigram算法、训练自定义分词器、与transformers集成

Transformers 模型库（结合源码）

PreTrainedModel/AutoModel架构、from_pretrained加载流程、generate解码、Trainer训练循环、量化

PEFT 参数高效微调库（结合源码）

LoRA/AdaLora/QLoRA数学原理、LoraLayer注入与合并、get_peft_model

TRL 对齐训练库（结合源码）

SFTTrainer/DPOTrainer/GRPOTrainer、DPO/GRPO/KTO数学原理、对齐训练流程

Accelerate 分布式训练库

Accelerator、DDP/FSDP/DeepSpeed集成、device_map自动分片、混合精度

Datasets 数据集库

load_dataset、map批量处理、流式加载(streaming)、Arrow零拷贝架构

bitsandbytes 量化库

LLM.int8()混合精度分解、NF4量化(QLoRA)、4bit/8bit推理与训练

DeepSpeed 分布式训练框架

ZeRO-1/2/3显存分析、ZeRO-Offload、激活检查点、HF Trainer集成

Megatron-Core 大模型训练框架

张量并行(Column/Row)、流水线并行(1F1B)、序列并行、上下文并行(Ring Attention)

Weights & Biases 实验追踪

wandb.init/log、Sweep超参搜索、Artifact版本管理、HF集成

MLflow ML生命周期管理

Tracking/Models/Registry/Projects、autolog、HuggingFace集成

vLLM 高性能推理引擎

PagedAttention原理、连续批处理、LLM类/SamplingParams、OpenAI兼容服务器

SGLang 结构化生成引擎

RadixAttention基数树KV cache复用、编程原语、约束解码

TensorRT-LLM 推理优化

计算图优化、内核融合、INT8/FP8量化、张量并行/流水线并行

llama.cpp 推理引擎

GGUF格式、Q4_0~Q6_K量化、CPU/GPU推理、mmap内存映射

llama-cpp-python Python绑定

Llama类、create_completion/chat、GBNF语法约束、嵌入提取、OpenAI服务器

FlashAttention 高效注意力

IO-awareness原理、在线Softmax、O(N²)→O(N)内存、分块计算推导

xFormers Transformer组件库

memory_efficient_attention、SwiGLU、RoPE旋转位置编码、稀疏注意力

FastAPI 异步Web框架

路由/Pydantic集成、SSE流式响应、中间件、LLM推理API服务

Gradio 交互式UI框架

Interface/Blocks、Chatbot组件、流式输出、HF Spaces部署

LangGraph Agent编排框架

StateGraph、节点/边/条件路由、持久化、Human-in-the-loop、多Agent

LlamaIndex RAG框架

Index/Retriever/QueryEngine、RAG管道抽象、Document/Node、Agent工具调用

DSPy 声明式Prompt编程

Signature、Module(Predict/ChainOfThought/ReAct)、Optimizer自动优化

AutoGen 多Agent对话框架

ConversableAgent、GroupChat、代码执行(本地/Docker)、v0.4 AgentChat

CrewAI 角色驱动协作

Agent/Task/Crew、Sequential/Hierarchical流程、Memory记忆系统

instructor 结构化输出

response_model、流式模式、Mode(TOOLS/JSON/MD_JSON)、Pydantic验证

Outlines 约束解码

generate.text/choice/regex/json/cfg、FSM驱动采样、logit偏置原理

smolagents 轻量Agent框架

CodeAgent/ToolCallingAgent、@tool装饰器、HfApiModel、代码沙箱

Ragas RAG评估框架

evaluate()、Faithfulness/AnswerRelevancy/ContextPrecision等指标、TestsetGenerator

DeepEval 评估测试框架

LLMTestCase、Hallucination/Bias/Toxicity等指标、GEval通用评估

LangSmith SDK 可观测性

@traceable追踪、evaluate()评估、数据集管理、trace→span层级

Arize Phoenix 开源可观测性

OpenTelemetry追踪、嵌入UMAP可视化、评估器、本地优先架构

Chroma 轻量向量数据库

PersistentClient、Collection CRUD、嵌入函数、元数据过滤、RAG存储

Qdrant 高性能向量数据库

HNSW索引、Filter过滤系统、Payload索引、量化、FastEmbed集成

Milvus 分布式向量数据库

IVF/PQ/HNSW索引、hybrid_search、存算分离架构、MilvusClient

pgvector PostgreSQL向量扩展

vector/halfvec/sparsevec类型、HNSW/IVFFlat索引、SQL距离操作符

portfolio

ARH — AI Research Helper Permalink

Open-source CLI-first research assistant agent for AI researchers. Tool use, three-layer memory, plan–execute safety, skill self-learning, multi-LLM fallback.

publications

FinVault: Benchmarking Financial Agent Safety in Execution-Grounded Environments

Published in arXiv preprint, 2026

The first execution-grounded security benchmark for LLM-based financial agents — 31 regulatory sandbox scenarios, 107 real-world vulnerabilities, 963 test cases. arXiv preprint.

VeriBRT: Plan-Guided, Evidence-Based Automated Bug Reproduction

Published in Manuscript, 2026

An LLM-based framework for generating bug-reproducing tests using a persistent Two-Axis Plan and reliability-aware evidence. Manuscript available online.

Reasoning-Visual Critical Token Fine-Tuning for Multimodal Reasoning

Published in Manuscript, 2026

RVCFT applies direct supervision selectively to response tokens that are both reasoning-relevant and sensitive to visual evidence. Manuscript available online.

talks

teaching