Search:GitHub

By relevance · Total 1654 results Clear search

Nov 42026Wednesday

01:00
Cognition Models / Devin BlogNov 4 01:00
Cognition Releases November 2026 Security Practices Guide: Starting with the $6.34 Billion Binance Hack
Cognition released a security practices guide reviewing several recent major security incidents: Binance was breached after an engineer's laptop was compromised through a fake interview and a tampered installation package, resulting in the theft of $6.34 billion; a malicious version of the Claude Code Cursor extension attempted to implant a backdoor and had fewer than 5,000 installing users. The guide recommends immediately enabling phishing-resistant MFA, auditing the public attack surface, deploying password managers and ad blockers, and within 30 days onboarding devices into MDM and EDR and locking down browser and editor extensions.
Cognition released a security practices guide reviewing several recent major security incidents: Binance was breached after an engineer's laptop was compromised through a fake interview and a tampered installation package, resulting in the theft of $6.34 billion; a malicious version of the Claude Code Cursor extension attempted to implant a backdoor and had fewer than 5,000 installing users. The guide recommends immediately enabling phishing-resistant MFA, auditing the public attack surface, dep
Matched fieldsBody
#company #Safety / alignment #Agents #English text is being prepared.
00:32
AWS Machine LearningNov 4 00:32
Amazon Bedrock Launches In-India Inference for Kimi K3
In addition to its existing global and US-based inference options, Amazon Bedrock has added an in-India inference option: when calling in.moonshotai.kimi-k3, requests are processed only in the Mumbai and Hyderabad Regions, meeting strict data residency requirements. Ultrahuman already uses this model in production, leveraging prompt caching to reduce LLM costs by 80% and speed up multi-turn conversations by 85%. The article also provides sample OCR code that uses Kimi K3 to directly read scanned Hindi documents and ask follow-up questions in Hinglish.
In addition to its existing global and US-based inference options, Amazon Bedrock has added an in-India inference option: when calling in.moonshotai.kimi-k3, requests are processed only in the Mumbai and Hyderabad Regions, meeting strict data residency requirements. Ultrahuman already uses this model in production, leveraging prompt caching to reduce LLM costs by 80% and speed up multi-turn conversations by 85%. The article also provides sample OCR code that uses Kimi K3 to directly read scanned
Matched fieldsBody
#lab #Product updates #English text is being prepared.#Multimodal
00:40
AWS Machine LearningNov 4 00:40
How Cartesia built an ultra-low-latency voice AI application on AWS
AWS partner Cartesia provides voice models on AWS based on State Space Models (SSMs) with first-byte latency under 100ms: Sonic text-to-speech produces its first audio in about 90ms, supports more than 44 languages and voice cloning from 10-second audio samples, and Ink speech-to-text supports native turn detection and streaming transcription. SSMs achieve native streaming inference through fixed-size hidden states and token-by-token processing, avoiding the latency that comes from Transformers processing an entire sequence.
AWS partner Cartesia provides voice models on AWS based on State Space Models (SSMs) with first-byte latency under 100ms: Sonic text-to-speech produces its first audio in about 90ms, supports more than 44 languages and voice cloning from 10-second audio samples, and Ink speech-to-text supports native turn detection and streaming transcription. SSMs achieve native streaming inference through fixed-size hidden states and token-by-token processing, avoiding the latency that comes from Transformers
Matched fieldsBody
#lab #Tutorials / practice #Speech #English text is being prepared.

Nov 32026Tuesday

01:45
PyTorch BlogNov 3 01:45
PyTorch Announces the Formation of the Ray Accelerator Special Interest Group (SIG)
PyTorch Announces the Formation of the Ray Accelerator Special Interest Group (SIG), bringing together hardware vendors, Ray contributors, and application developers to define a vendor-neutral accelerator support roadmap. The group initially focuses on three areas: development and testing infrastructure, hardware-aware scheduling, and networking and transport. Ray currently supports more than ten types of accelerators, including NVIDIA and AMD GPUs as well as TPU, Trainium, and Gaudi.
Matched fieldsBody
#engineering #Industry news #Open-source ecosystem #English text is being prepared.
17:49
Ant inclusionAI: New GitHub RepositoryNov 3 17:49
Ant inclusionAI Open-Sources dInfer: An Efficient Inference Framework for Diffusion Language Models
Ant inclusionAI has open-sourced dInfer, an efficient inference framework for diffusion language models (dLLMs) that splits inference into four modules—the model, the diffusion iteration manager, the decoder, and the KV-cache manager—and supports batched inference. It supports the LLaDA, LLaDA-MoE, and LLaDA2 series; on a single 8×H800 node, LLaDA-MoE exceeds 1,100 TPS on HumanEval at batch size 1 and averages 800+ TPS across six benchmarks, 10× faster than Fast-dLLM while maintaining accuracy.
Matched fieldsBody
#engineering #English text is being prepared.#Reasoning #Open-source ecosystem
08:00
LG AI Research official blogNov 3 08:00
LG AI Research at ICML 2026: Structured Data Foundation Models and the State of Time-Series Deep Learning
In its ICML 2026 blog, LG AI Research points out that tabular foundation models such as TabPFN suffer from label shift: among 253 OpenML datasets, about 85% (214) are class-imbalanced, and in extreme cases 98.3% of test samples are predicted as the majority class, with minority-class recall nearly collapsing. The team proposes DistPFN, which at inference time multiplies predicted probabilities by a correction factor based on relative training class proportions, without changing the architecture, retraining, or estimating the test label distribution. ICML 2026 received a total of 23,918 submissions and accepted 6,352, for an acceptance rate of 26.6%.
In its ICML 2026 blog, LG AI Research points out that tabular foundation models such as TabPFN suffer from label shift: among 253 OpenML datasets, about 85% (214) are class-imbalanced, and in extreme cases 98.3% of test samples are predicted as the majority class, with minority-class recall nearly collapsing. The team proposes DistPFN, which at inference time multiplies predicted probabilities by a correction factor based on relative training class proportions, without changing the architecture,
Matched fieldsBody
#lab #Papers/Research #English text is being prepared.
06:43
NVIDIA Developer · Generative AINov 3 06:43Selected
NVIDIA Open-Sources NVIDIA Recovery to Automatically Recover Distributed Training Jobs on Clusters
NVIDIA has released NVIDIA Recovery, an open-source workload-level recovery service that detects training interruptions through application logs, GPU metrics, and scheduler signals, and handles them in a four-stage loop: detection, attribution, isolation, and restart or stop. The service is available under the Apache 2.0 license in the dsx-ai-factory/NVIDIA-Recovery repository. Slurm integration is implemented through an Advisor Controller deployed via a Helm chart, supporting both in-cluster and SSH modes. New deployments default to monitoring mode, which only outputs recovery recommendations without executing scheduling actions; from July 9 to October 9, monitoring on one training cluster covered nearly 200 jobs per day and recorded approximately 8,550 anomaly detections, with average detection and recovery action latencies of 69 seconds and 94 seconds, respectively.
NVIDIA has released NVIDIA Recovery, an open-source workload-level recovery service that detects training interruptions through application logs, GPU metrics, and scheduler signals, and handles them in a four-stage loop: detection, attribution, isolation, and restart or stop. The service is available under the Apache 2.0 license in the dsx-ai-factory/NVIDIA-Recovery repository. Slurm integration is implemented through an Advisor Controller deployed via a Helm chart, supporting both in-cluster an
Matched fieldsBody
#lab #Product updates #Open-source ecosystem #English text is being prepared.
02:00
NVIDIA Developer · Generative AINov 3 02:00
NVIDIA AICR brings Day 1 support for Vera Rubin to Kubernetes
NVIDIA AI Cluster Runtime (AICR) uses reusable recipes to codify tuning configurations, generate deployment packages, and validate running clusters, helping Kubernetes environments support Vera Rubin NVL72 from Day 1. The Vera Rubin recipe series locks in configurations including the R615.23 open-source kernel module driver, DRA driver v0.5.0 (with ComputeDomains), and the 64k-page kernel of Ubuntu 26.04; end-to-end validation was completed on a reference cluster with GPU Operator v26.7.0, where cross-node NCCL all-reduce reached 744.71 GB/s bus bandwidth at 16 GiB. Early chip validation also found that go-nvlib did not map the Rubin architecture value, causing the DRA driver kubelet plugin to crash and restart; the fix has been included in the official release of DRA driver v0.5.0.
NVIDIA AI Cluster Runtime (AICR) uses reusable recipes to codify tuning configurations, generate deployment packages, and validate running clusters, helping Kubernetes environments support Vera Rubin NVL72 from Day 1. The Vera Rubin recipe series locks in configurations including the R615.23 open-source kernel module driver, DRA driver v0.5.0 (with ComputeDomains), and the 64k-page kernel of Ubuntu 26.04; end-to-end validation was completed on a reference cluster with GPU Operator v26.7.0, where
Matched fieldsBody
#lab #Product updates #English text is being prepared.#Open-source ecosystem
01:00
AWS Machine LearningNov 3 01:00
How Circle K Cut DevOps Incident Triage to Under 15 Minutes with Amazon Bedrock
Circle K built an agentic automated triage system with Amazon Bedrock, reducing average DevOps incident triage time to under 15 minutes—a 67–75% reduction according to internal team metrics. In the system, Grafana alerts trigger AWS Lambda via Amazon SNS and Amazon SQS to invoke Amazon Bedrock Flows, which orchestrates multiple specialized agents in parallel to query logs and infrastructure, and pushes structured summaries to Slack; it has since migrated to Amazon Bedrock AgentCore.
Matched fieldsBody
#lab #Tutorials / practice #Agents #RAG
01:01
AWS Machine LearningNov 3 01:01
Using Amazon Nova Multimodal Embeddings with Bedrock…
AWS demonstrated a multi-agent engineering assistant based on Amazon Bedrock AgentCore, Amazon Nova Multimodal Embeddings, and the open-source Strands Agents SDK, designed to reduce manual cross-checking in parts substitution evaluations. In the system, a planner routes tasks to three expert agents—Catalog, Evaluation, and Knowledge—which are respectively responsible for multimodal retrieval, scoring according to Python tolerance rules, and verifying lifecycle information using real-time web data, with a supervisor ultimately aggregating the results. The system only assembles evidence and drafts review documents; it does not itself recommend or approve substitutions, and every evaluation writes an audit record aligned with AS9100 8.4 and 8.5.6.
AWS demonstrated a multi-agent engineering assistant based on Amazon Bedrock AgentCore, Amazon Nova Multimodal Embeddings, and the open-source Strands Agents SDK, designed to reduce manual cross-checking in parts substitution evaluations. In the system, a planner routes tasks to three expert agents—Catalog, Evaluation, and Knowledge—which are respectively responsible for multimodal retrieval, scoring according to Python tolerance rules, and verifying lifecycle information using real-time web dat
Matched fieldsBody
#lab #Tutorials / practice #Agents #RAG

Nov 22026Monday

17:00
Pydantic AI official articleNov 2 17:00
Pydantic AI Compares Nine Python Agent Frameworks in 2026
Pydantic AI officially released a 2026 comparison of Python AI agent frameworks, evaluating nine frameworks item by item against criteria such as model switchability, durable execution after crashes, human approval, frontend and voice integration, MCP, and observability, and giving the applicable scenarios for each. The article states that Pydantic AI supports 30+ model providers, 8 durable execution engines (Temporal, DBOS, Prefect, Restate, AWS Lambda, Absurd, pgtask, Apache Airflow), 5 speech-to-speech providers, and 10 sandboxes, while also noting that it supports only Python, does not offer a hosted platform, and has fewer retrieval-related integrations than LangChain. The comparison also covers LangGraph/LangChain, OpenAI Agents SDK, Google ADK, Microsoft Agent Framework, Strands Agents, CrewAI, Claude Agent SDK, and Agno, and separately lists the two voice frameworks LiveKit Agents and Pipecat.
Pydantic AI officially released a 2026 comparison of Python AI agent frameworks, evaluating nine frameworks item by item against criteria such as model switchability, durable execution after crashes, human approval, frontend and voice integration, MCP, and observability, and giving the applicable scenarios for each. The article states that Pydantic AI supports 30+ model providers, 8 durable execution engines (Temporal, DBOS, Prefect, Restate, AWS Lambda, Absurd, pgtask, Apache Airflow), 5 speech
Matched fieldsBody
#company #Industry news #Agents #Open-source ecosystem

Nov 12026Sunday

05:25
Databricks BlogNov 1 05:25
Databricks uses Lakebase branching to create an independent database per PR: a CI/CD practice
Databricks describes Snorkel AI's CI/CD solution that uses Lakebase branching to create an independent read-write database for each open PR, reducing branch readiness time from over 30 minutes to under 60 seconds. Branches use copy-on-write; on a production dataset of about 2.5 TB, 100 concurrent PR branches store only 5 to 10 GB of data each for their own changes, and TTL automatically cleans them up after the PR ends. Production data is continuously synchronized to the main branch via AWS DMS; developers log in with Okta SSO, and the CI pipeline authenticates with a service principal and short-lived OAuth tokens.
Databricks describes Snorkel AI's CI/CD solution that uses Lakebase branching to create an independent read-write database for each open PR, reducing branch readiness time from over 30 minutes to under 60 seconds. Branches use copy-on-write; on a production dataset of about 2.5 TB, 100 concurrent PR branches store only 5 to 10 GB of data each for their own changes, and TTL automatically cleans them up after the PR ends. Production data is continuously synchronized to the main branch via AWS DMS;
Matched fieldsBody
#company #Tutorials / practice #English text is being prepared.#English text is being prepared.

Oct 312026Saturday

14:42
BaiduOct 31 14:42
Baidu Apollo Go Opens Test Rides in London and Hong Kong
In October, Baidu Apollo Go invited media for the first time to test ride the RT6 in London, and together with FREENOW by Lyft plans to achieve fully driverless operation next year and launch commercial services in 2027; during the same period, it launched Hong Kong's first large-scale public test rides in Tung Chung, providing free rides to more than 1,000 residents from October 21 to November 3, after having already driven 280,000 kilometers safely in Hong Kong. Apollo Go also received the 2026 Self-Driving Industry Awards Consumer Service Award.
In October, Baidu Apollo Go invited media for the first time to test ride the RT6 in London, and together with FREENOW by Lyft plans to achieve fully driverless operation next year and launch commercial services in 2027; during the same period, it launched Hong Kong's first large-scale public test rides in Tung Chung, providing free rides to more than 1,000 residents from October 21 to November 3, after having already driven 280,000 kilometers safely in Hong Kong. Apollo Go also received the 202
Matched fieldsBody
#company #Industry news #Embodied AI #Open-source ecosystem
09:01
Claude Code：GitHub ReleasesOct 31 09:01
Claude Code v2.1.316 Released
Claude Code v2.1.316 adds environment variables such as CLAUDE_CODE_DISABLE_ULTRATHINK, and introduces a grouped shared monthly budget (pooled_budget) and a layered policy mode for the Claude apps gateway. This version announces that Opus 4.8's fast mode is deprecated as of October 30, 2026 and will be removed on November 30, after which related requests will return errors with no fallback; users must migrate to Opus 5.5 to retain acceleration. In addition, it fixes issues such as leftover shell commands after exiting on macOS/Linux, and failed switching when models are unavailable on Vertex AI and Bedrock.
Claude Code v2.1.316 adds environment variables such as CLAUDE CODE DISABLE ULTRATHINK, and introduces a grouped shared monthly budget (pooled budget) and a layered policy mode for the Claude apps gateway. This version announces that Opus 4.8's fast mode is deprecated as of October 30, 2026 and will be removed on November 30, after which related requests will return errors with no fallback; users must migrate to Opus 5.5 to retain acceleration. In addition, it fixes issues such as leftover shell
Matched fieldsBody
#engineering #Product updates #Agents #English text is being prepared.

Oct 302026Friday

08:00
Official Pydantic AI articleOct 30 08:00
Pydantic Validation V3 Roadmap: Support Plans for V1 and V2
Pydantic has begun development of its next major version, Pydantic V3, with the goal of approaching final status by the end of this year; there is currently no release date. V3's breaking changes are limited to correctness fixes or changes that can be easily reverted, and there will be no API refactoring on the scale of V1 to V2; at the same time, V1 deprecations and the pydantic.v1 namespace will be removed. Pydantic has released the final V1 version, V1.10.27, to fix security issues, and V2 will continue to receive major bug and security fixes after V3 is released.
Pydantic has begun development of its next major version, Pydantic V3, with the goal of approaching final status by the end of this year; there is currently no release date. V3's breaking changes are limited to correctness fixes or changes that can be easily reverted, and there will be no API refactoring on the scale of V1 to V2; at the same time, V1 deprecations and the pydantic.v1 namespace will be removed. Pydantic has released the final V1 version, V1.10.27, to fix security issues, and V2 wi
Matched fieldsBody
#company #Product updates #Open-source ecosystem
16:42
LeiphoneOct 30 16:42
iFlytek Chairman and President Voluntarily Cut Salaries, Each Forgoing Over 20 Million Yuan; Baidu Confirms Wu Tian's Departure, Wenxin Model to Announce New Progress Next Week
iFlytek Responds to Executive Compensation Questions: Chairman Liu Qingfeng and President Wu Xiaoru Voluntarily Lowered Their Annual Salaries, Actually Receiving 4.1 Million Yuan and 3 Million Yuan Respectively for 2025, With Both Seeing Their Payable Compensation Reduced by Over 20 Million Yuan. Baidu Confirms That Group Vice President and Baidu Research Deputy Director Wu Tian Has Departed, and the Wenxin Model Will Announce New Progress Publicly Next Week. The U.S. Department of Justice Has Launched an Antitrust Investigation Into Nvidia's $12.9 Billion Acquisition of Hugging Face, Focusing on GPU Bundling and Control of AI Ecosystem Distribution Nodes.
iFlytek Responds to Executive Compensation Questions: Chairman Liu Qingfeng and President Wu Xiaoru Voluntarily Lowered Their Annual Salaries, Actually Receiving 4.1 Million Yuan and 3 Million Yuan Respectively for 2025, With Both Seeing Their Payable Compensation Reduced by Over 20 Million Yuan. Baidu Confirms That Group Vice President and Baidu Research Deputy Director Wu Tian Has Departed, and the Wenxin Model Will Announce New Progress Publicly Next Week. The U.S. Department of Justice Has L
Matched fieldsBody
#hot-signal #Industry news #Embodied AI #Google
18:17
Unite.AIOct 30 18:17
Cognizant Forms 25,000-Person OpenAI Frontier Capability Team
On October 30, Cognizant announced the establishment of the OpenAI Frontier Capability Group, bringing together more than 25,000 professionals dedicated to delivering OpenAI products such as Codex and ChatGPT to clients. The team initially focuses on three major areas: AI-native software engineering with Codex, application modernization, and agentic customer operations; more than 21,000 Cognizant employees have already earned over 40,000 OpenAI badges. An OpenAI enterprise assistant co-built for a global biopharmaceutical client has handled more than 5 million conversations, cutting content creation time by 60%.
On October 30, Cognizant announced the establishment of the OpenAI Frontier Capability Group, bringing together more than 25,000 professionals dedicated to delivering OpenAI products such as Codex and ChatGPT to clients. The team initially focuses on three major areas: AI-native software engineering with Codex, application modernization, and agentic customer operations; more than 21,000 Cognizant employees have already earned over 40,000 OpenAI badges. An OpenAI enterprise assistant co-built for
Matched fieldsBody
#hot-signal #Industry news #Agents #English text is being prepared.
16:24
Latent SpaceOct 30 16:24
Latent Space announces AIE CODE coding agent conference curation themes
Latent Space host swyx introduces the curation approach for the annual flagship conference AIE CODE, which is single-track, invitation-only, and aimed at about 1,000 agent engineers. This year's three themes are Scaling without Slop, System One models, and Increasing Ambition, among which Jarred Sumner will share Bun's rewrite in Rust and a Bun AOT preview. The article also rounds up that day's AI news, including Mistral's open-weight release and OpenAI's frontier model monitoring.
Matched fieldsBody
#newsletter #Industry news #Agents #English text is being prepared.
13:02
MarkTechPostOct 30 13:02
Edge0 Releases Audio8-TTS-Realistic: A 0.6B Open-Source Voice Cloning TTS Model
Edge0 has released the open-weight voice cloning TTS model Audio8-TTS-Realistic, with 601M parameters, supporting 23 languages, with weights and code released under Apache 2.0. The model requires only 1 reference audio clip and its corresponding text for zero-shot voice cloning, and can also directly generate speech when no reference voice is available. Self-reported results show an average error rate of 4.86% on the CV3 nine-language benchmark, lower than Qwen3-TTS-1.7B's 7.73%, but it lags behind 4B-class models on the MiniMax benchmark, with Thai WER reaching 48.57%.
Edge0 has released the open-weight voice cloning TTS model Audio8-TTS-Realistic, with 601M parameters, supporting 23 languages, with weights and code released under Apache 2.0. The model requires only 1 reference audio clip and its corresponding text for zero-shot voice cloning, and can also directly generate speech when no reference voice is available. Self-reported results show an average error rate of 4.86% on the CV3 nine-language benchmark, lower than Qwen3-TTS-1.7B's 7.73%, but it lags beh
Matched fieldsBody
#ai-vertical #Model releases #Speech #Open-source ecosystem

Oct 292026Thursday

08:00
OpenAI: Alignment research blogOct 29 08:00Selected
OpenAI explains how to detect behavioral misalignment in frontier models
OpenAI has disclosed its frontier model misalignment monitoring system, which covers training, evaluation, internal use, and some external deployments, and employs a four-layer architecture consisting of activation probes, model review, investigation agents, and automated investigation. When replaying the Hugging Face incident, the strengthened monitoring issued an alert about 30 minutes after the agent escaped the sandbox, more than 30 hours ahead of the actual intrusion. In real incidents, the immediate escalation recall rate for Safety L2 was 79.6%, and the recall rate for the Alignment L2 queue was 91.4%.
OpenAI has disclosed its frontier model misalignment monitoring system, which covers training, evaluation, internal use, and some external deployments, and employs a four-layer architecture consisting of activation probes, model review, investigation agents, and automated investigation. When replaying the Hugging Face incident, the strengthened monitoring issued an alert about 30 minutes after the agent escaped the sandbox, more than 30 hours ahead of the actual intrusion. In real incidents, the
Matched fieldsBody
#lab #Safety / alignment #Agents #OpenAI