41554 blogs · [ { "id": "01a087a3-49df-70d7-9a69-b8024041062c", "title": "Harnesses for Generalisation", "url": "https://mdda.net/blog/research/talks/harnesses-for-generalisation", "published_at": "2026-08-06T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80240ca9aea", "title": "GDG I/O Extended Singapore : New Gemma Models", "url": "https://mdda.net/blog/research/talks/gemma-12B-and-diffusion", "published_at": "2026-06-27T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80241c315e9", "title": "Local Models & Self-Evolving Harnesses", "url": "https://mdda.net/blog/research/talks/own-your-ai", "published_at": "2026-06-25T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80241e03a4e", "title": "DSTA BrainHack event - Invited Talk", "url": "https://mdda.net/blog/research/talks/anyone-can-cook", "published_at": "2026-06-10T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802421a1d59", "title": "External Keypad for OBS Controls", "url": "https://mdda.net/blog/tech/misc/external-keypad-for-OBS-controls", "published_at": "2026-05-28T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80242563c9e", "title": "AI Engineer Gemini Event", "url": "https://mdda.net/blog/research/talks/gemini-flash-tts", "published_at": "2026-05-13T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024259d40d", "title": "Running ~30B MoE models on an old GTX 1080", "url": "https://mdda.net/blog/tech/dl/llama-cpp-moe-on-an-old-gtx-1080", "published_at": "2026-05-11T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80242acce44", "title": "Robots : Reasoning and Acting", "url": "https://mdda.net/blog/research/talks/robot-world-models", "published_at": "2026-04-29T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80242ea06e6", "title": "Personal Agents - Cloud and Edge", "url": "https://mdda.net/blog/research/talks/personal-agents", "published_at": "2026-02-26T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024306a31f", "title": "Signals Between the Tokens", "url": "https://mdda.net/blog/research/ideas/signals-between-the-tokens", "published_at": "2026-02-24T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80243cda0cf", "title": "Learning to Think, Not Learning to Solve", "url": "https://mdda.net/blog/research/ideas/learning-to-think-not-learning-to-solve", "published_at": "2026-02-17T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024481243d", "title": "NeurIPS Roundup", "url": "https://mdda.net/blog/research/talks/NeurIPS-roundup", "published_at": "2026-01-15T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80244e7d19a", "title": "Sizing gemma3 for TPUv5-n", "url": "https://mdda.net/blog/tech/misc/gemma3-training-sizing-on-TPUv5", "published_at": "2025-12-16T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802458f4194", "title": "Practical TPU VM on GCP", "url": "https://mdda.net/blog/tech/misc/practical-TPU-VM-on-GCP", "published_at": "2025-12-15T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802459cf493", "title": "Reinforcement Learning for Long-Horizon Multi-Turn Search Agents", "url": "https://mdda.net/blog/research/papers/NeurIPS-Workshop-RL-for-MultiTurn-RAG", "published_at": "2025-12-06T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802467880e2", "title": "Evolving Reasoning", "url": "https://mdda.net/blog/research/talks/evolving-reasoning", "published_at": "2025-11-26T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024685c665", "title": "DevFest SG Talk: Lessons from Reinforcement Learning", "url": "https://mdda.net/blog/research/talks/lessons-from-rl-at-singapore-devfest", "published_at": "2025-11-22T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80247201e13", "title": "Neural Assets and World Models", "url": "https://mdda.net/blog/research/talks/neural-assets-world-models", "published_at": "2025-10-15T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80248056f6c", "title": "IPhO Gold using Agentic Gemini", "url": "https://mdda.net/blog/research/talks/agentic-gemini-for-ipho", "published_at": "2025-09-25T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802489cb69b", "title": "Agent Efficiency, Memory and Confidence", "url": "https://mdda.net/blog/research/talks/agent-efficiency-memory-confidence", "published_at": "2025-08-27T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80248ada7af", "title": "A Reasoning-Based Approach to Cryptic Crossword Clue Solving", "url": "https://mdda.net/blog/research/papers/ICML-2025-Conference_reasoning-based-approach-to-cryptic-crossword-clue-solving", "published_at": "2025-07-16T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80249155aa1", "title": "GPU Kernel Scientist : An LLM-Driven Framework for Iterative Kernel Optimization", "url": "https://mdda.net/blog/research/papers/gpu-kernel-scientist", "published_at": "2025-07-03T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024a003510", "title": "Self-Improving Agents", "url": "https://mdda.net/blog/research/talks/self-improving-agents", "published_at": "2025-06-19T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024a7c2981", "title": "Evolving GPU Kernels", "url": "https://mdda.net/blog/research/talks/evolving-gpu-kernels", "published_at": "2025-05-22T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024b2207da", "title": "Latent Space Reasoning", "url": "https://mdda.net/blog/research/talks/latent-space-reasoning", "published_at": "2025-04-30T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024b91111a", "title": "Diffusion LLMs", "url": "https://mdda.net/blog/research/talks/diffusion-llms", "published_at": "2025-03-25T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024bb7141e", "title": "Getting to Aha (notes)", "url": "https://mdda.net/blog/tech/dl/getting-to-aha-sprint", "published_at": "2025-02-28T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024c942a43", "title": "Getting to Aha", "url": "https://mdda.net/blog/research/talks/getting-to-aha", "published_at": "2025-02-25T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024ce8dd8d", "title": "NeurIPS recap and SOTA", "url": "https://mdda.net/blog/research/talks/NeurIPS-recap-and-SOTA", "published_at": "2025-01-22T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024dd1d5c7", "title": "System 2 - Training for Test-time compute", "url": "https://mdda.net/blog/research/talks/rl-for-llms-o1", "published_at": "2024-09-26T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024ea08336", "title": "Slack to Mattermost on New Fedora server", "url": "https://mdda.net/blog/tech/misc/slack-to-mattermost-on-new-fedora-server", "published_at": "2024-09-20T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024f01cb0e", "title": "o1 - Are the curves awesome?", "url": "https://mdda.net/blog/research/ideas/o1-log-linear-analysis", "published_at": "2024-09-15T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024fbfdcc4", "title": "Fedora 40 XFCE Live modifications", "url": "https://mdda.net/blog/tech/dl/fedora-40-xfce-live-modifications", "published_at": "2024-09-14T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8024fdff7d2", "title": "Install cuda-12.6 on Fedora 40", "url": "https://mdda.net/blog/tech/dl/install-fedora-40-nvidia", "published_at": "2024-09-14T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802507c0d71", "title": "Upgrade to Fedora 40 with cuda-12.4", "url": "https://mdda.net/blog/tech/dl/upgrade-fedora-40-with-cuda", "published_at": "2024-08-30T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025159f7fd", "title": "Proving that Cryptic Crossword Clue Answers are Correct", "url": "https://mdda.net/blog/research/papers/ICML-2024-LLMs-and-Cognition-proving-cryptic-crossword-clue-answers-are-correct", "published_at": "2024-07-27T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80251f3b4d3", "title": "Small LM Agents", "url": "https://mdda.net/blog/research/talks/small-lm-agents", "published_at": "2024-07-17T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802526e42d2", "title": "Cryptic Proof Checking", "url": "https://mdda.net/blog/research/talks/cryptic-proof-formalisation", "published_at": "2024-06-18T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80252835850", "title": "Upgrade to Fedora 38 with cuda-12.4", "url": "https://mdda.net/blog/tech/dl/upgrade-fedora-38-for-cuda-12.4", "published_at": "2024-05-10T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802531c3486", "title": "Long Context Transformers", "url": "https://mdda.net/blog/research/talks/long-context-transformers", "published_at": "2024-04-25T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80253712e45", "title": "DSPy - Gemma Sprint", "url": "https://mdda.net/blog/tech/dl/DSPy-gemma", "published_at": "2024-04-08T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025374b1e3", "title": "Gemini 1.5 One Million Sprint", "url": "https://mdda.net/blog/tech/dl/gemini-one-million", "published_at": "2024-03-23T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80254450710", "title": "DSPy - Gemini and Gemma", "url": "https://mdda.net/blog/research/talks/DSPy-gemini-and-gemma", "published_at": "2024-03-14T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025540c8c4", "title": "Gemini 1 million", "url": "https://mdda.net/blog/research/talks/gemini-1-million", "published_at": "2024-02-20T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80255f08aff", "title": "Research from NeurIPS", "url": "https://mdda.net/blog/research/talks/research-from-NeurIPS", "published_at": "2024-01-18T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80256ce08cf", "title": "Enable bluetooth management via GUI on Fedora", "url": "https://mdda.net/blog/tech/misc/bluetooth-gui-permissions-fedora", "published_at": "2023-12-04T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802577cc92b", "title": "DevFest SG Talk: No-code Prototyping with Flowise", "url": "https://mdda.net/blog/research/talks/flowise-nocode-langchain-at-singapore-devfest", "published_at": "2023-11-18T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80257e373d1", "title": "What's Next in RAG", "url": "https://mdda.net/blog/research/talks/what-is-next-in-rag", "published_at": "2023-11-16T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025816cb30", "title": "Converting a Presentation Transcript to Speaker Notes", "url": "https://mdda.net/blog/tech/dl/speaker-notes-from-transcript-using-makersuite-palm", "published_at": "2023-10-27T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80258a58cd8", "title": "LLMs and Reasoning", "url": "https://mdda.net/blog/research/talks/llms-and-reasoning", "published_at": "2023-09-14T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802592784f8", "title": "Keras Core", "url": "https://mdda.net/blog/research/talks/keras-core", "published_at": "2023-09-09T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80259a23b29", "title": "Upgrade to Fedora 38 including Nvidia, etc", "url": "https://mdda.net/blog/tech/dl/upgrade-to-fedora-38-including-nvidia-etc", "published_at": "2023-07-28T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80259b24fee", "title": "Open-Endedness and Large Language Models", "url": "https://mdda.net/blog/research/talks/open-endedness-and-large-language-models", "published_at": "2023-07-27T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80259f813f5", "title": "Self-Improvement (for Language Models)", "url": "https://mdda.net/blog/research/talks/self-improvement-for-language-models-", "published_at": "2023-07-13T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80259fad097", "title": "Code and Reasoning with LLMs", "url": "https://mdda.net/blog/research/talks/code-and-reasoning-with-llms", "published_at": "2023-06-26T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025af67c43", "title": "MediaPipe with a bit of Bard", "url": "https://mdda.net/blog/research/talks/mediapipe-with-a-bit-of-bard", "published_at": "2023-05-28T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025b7444af", "title": "Images+LLMs : Advances in Multimodal Models", "url": "https://mdda.net/blog/research/talks/images-llms-advances-in-multimodal-models", "published_at": "2023-05-25T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025c121bf5", "title": "Interacting with LLMs in a post-chatGPT world", "url": "https://mdda.net/blog/research/talks/interacting-with-llms-in-a-post-chatgpt-world", "published_at": "2023-04-20T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025c764566", "title": "Cloud Platforms & Other Super Fun Stuff", "url": "https://mdda.net/blog/research/talks/cloud-platforms-other-super-fun-stuff", "published_at": "2023-03-13T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025c7883f1", "title": "News You Can Use - Practical Stuff", "url": "https://mdda.net/blog/research/talks/news-you-can-use-practical-stuff", "published_at": "2023-02-23T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025cb15124", "title": "Voices In, Voices Out - ASR, TTS", "url": "https://mdda.net/blog/research/talks/voices-in-voices-out-asr-tts", "published_at": "2023-01-19T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025d0eea0b", "title": "Prompting, Instructions and the Future of Large Language Models", "url": "https://mdda.net/blog/research/talks/prompting-instructions-and-the-future-of-large-language-models", "published_at": "2022-11-29T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025d3e26c0", "title": "World of Diffusion : Theory, Shortcuts & Future", "url": "https://mdda.net/blog/research/talks/world-of-diffusion-theory-shortcuts-future", "published_at": "2022-09-15T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025dcf284b", "title": "Running MineRL within a Container", "url": "https://mdda.net/blog/research/ideas/running-minerl-within-a-container", "published_at": "2022-09-13T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025e0c99f2", "title": "Containers within a GCP host server, with Nvidia pass-through", "url": "https://mdda.net/blog/tech/dl/containers-within-a-gcp-host-server-with-nvidia-pass-through", "published_at": "2022-09-12T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025e3f0ab2", "title": "GCP VM with Nvidia set-up and Convenience Scripts", "url": "https://mdda.net/blog/tech/dl/gcp-vm-with-nvidia-set-up-and-convenience-scripts", "published_at": "2022-09-02T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025f0c6608", "title": "What’s New in AI & Machine Learning", "url": "https://mdda.net/blog/research/talks/what-s-new-in-ai-machine-learning", "published_at": "2022-08-18T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8025f95983f", "title": "RETRO vs LaMDA - Clash of the Transformers", "url": "https://mdda.net/blog/research/talks/retro-vs-lamda-clash-of-the-transformers", "published_at": "2022-01-25T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802600cd329", "title": "Text Similarity for NLP", "url": "https://mdda.net/blog/research/talks/text-similarity-for-nlp", "published_at": "2022-01-07T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80260e76b0c", "title": "Back to Basics 03", "url": "https://mdda.net/blog/research/talks/back-to-basics-03-run2", "published_at": "2021-12-16T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802610d680d", "title": "Back to Basics 02", "url": "https://mdda.net/blog/research/talks/back-to-basics-02-run2", "published_at": "2021-12-09T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80261cdb007", "title": "Back to Basics 01", "url": "https://mdda.net/blog/research/talks/back-to-basics-01-run2", "published_at": "2021-12-02T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80262aef6cc", "title": "GPT : Beams, Sampling & more", "url": "https://mdda.net/blog/research/talks/gpt-beams-sampling-amp-more", "published_at": "2021-11-25T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80262af54ed", "title": "Investigating Prompt Engineering in Diffusion Models", "url": "https://mdda.net/blog/research/papers/arxiv-investigating-prompt-engineering-in-diffusion-models", "published_at": "2021-11-21T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8026302b56d", "title": "Using a GCP VM like a 'local' Deep Learning machine", "url": "https://mdda.net/blog/tech/dl/using-a-gcp-vm-like-a-local-deep-learning-machine", "published_at": "2021-10-10T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802639654ee", "title": "Adding Firewalled Jupyter to a GCP VM", "url": "https://mdda.net/blog/tech/dl/adding-firewalled-jupyter-to-a-gcp-vm", "published_at": "2021-10-07T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802640f8928", "title": "Building a reusable Deep Learning VM on Google Cloud", "url": "https://mdda.net/blog/tech/dl/building-a-reusable-deep-learning-vm-on-google-cloud", "published_at": "2021-10-05T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80264511401", "title": "Nvidia (11.3) for TensorFlow & PyTorch while upgrading Fedora 34", "url": "https://mdda.net/blog/tech/dl/nvidia-11-3-for-tensorflow-pytorch-while-upgrading-fedora-34", "published_at": "2021-09-04T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80264d41200", "title": "PerceiverIO", "url": "https://mdda.net/blog/research/talks/perceiverio", "published_at": "2021-08-26T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802659ea17d", "title": "Exploring Different Approaches for Multilingual Tasks", "url": "https://mdda.net/blog/research/papers/ACL-IJCNLP-2021-CASE-exploring-approaches-for-multilingual-tasks", "published_at": "2021-08-05T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802660cf3d1", "title": "Back to Basics 03", "url": "https://mdda.net/blog/research/talks/back-to-basics-03", "published_at": "2021-07-20T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802670848f7", "title": "Back to Basics 02", "url": "https://mdda.net/blog/research/talks/back-to-basics-02", "published_at": "2021-07-13T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802678070e5", "title": "Back to Basics 01", "url": "https://mdda.net/blog/research/talks/back-to-basics-01", "published_at": "2021-07-06T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80268105e53", "title": "Multi-Hop Inference Explanation Regeneration by Matching Expert Ratings", "url": "https://mdda.net/blog/research/papers/NAACL-2021-Textgraphs-multihop-explanation-regeneration-expert-ratings", "published_at": "2021-06-11T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80268cbfecd", "title": "Positioning Yourself for the Future: AI from the Backend", "url": "https://mdda.net/blog/research/talks/positioning-yourself-for-the-future-ai-from-the-backend", "published_at": "2021-04-10T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802692b164e", "title": "TensorFlow Everywhere", "url": "https://mdda.net/blog/research/talks/tensorflow-everywhere", "published_at": "2021-02-18T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b80269590584", "title": "LIT : LSTM-Interleaved Transformer for Multi-Hop Explanation Ranking", "url": "https://mdda.net/blog/research/papers/COLING-2020-Textgraphs-lstm-interleaved-transformer-multihop-explanation-ranking", "published_at": "2020-12-13T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802698687ac", "title": "Where the jobs are in AI, and Why", "url": "https://mdda.net/blog/research/talks/where-the-jobs-are-in-ai-and-why", "published_at": "2020-09-12T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8026a706275", "title": "TensorFlow 2.3 from source on Fedora 32", "url": "https://mdda.net/blog/research/ideas/tensorflow-2-3-from-source-on-fedora-32", "published_at": "2020-07-29T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8026aa49a71", "title": "Deep Learning State of the Art", "url": "https://mdda.net/blog/research/talks/deep-learning-state-of-the-art", "published_at": "2020-02-06T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8026b3e8f2d", "title": "TensorBoard Lite", "url": "https://mdda.net/blog/research/talks/tensorboard-lite", "published_at": "2019-12-19T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8026c0fa1db", "title": "TensorFlow 2.1 from source on Fedora 30", "url": "https://mdda.net/blog/research/ideas/tensorflow-2-1-from-source-on-fedora-30", "published_at": "2019-12-04T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8026c3ad792", "title": "Faces in the Browser using TensorFlow.js", "url": "https://mdda.net/blog/research/talks/faces-in-the-browser-using-tensorflow-js", "published_at": "2019-11-30T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8026d37f4f0", "title": "Neural Structured Learning", "url": "https://mdda.net/blog/research/talks/neural-structured-learning", "published_at": "2019-11-12T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8026daf6b23", "title": "Nvidia through docker on Fedora 30", "url": "https://mdda.net/blog/tech/dl/nvidia-through-docker-on-fedora-30", "published_at": "2019-11-12T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8026e1fc36d", "title": "Language Model Assisted Explanation Generation", "url": "https://mdda.net/blog/research/papers/EMNLP-IJCNLP-2019-Textgraphs-llm-assisted-explanation-generation", "published_at": "2019-11-04T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8026e7f0d50", "title": "Paraphrasing with Large Language Models", "url": "https://mdda.net/blog/research/papers/EMNLP-IJCNLP-2019-WNGT-paraphrasing-with-large-language-models", "published_at": "2019-11-04T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8026f26390a", "title": "Unsupervised Natural Question Answering with a Small Model", "url": "https://mdda.net/blog/research/papers/EMNLP-IJCNLP-2019-FEVER_unsupervised-natural-question-answering-with-small-model", "published_at": "2019-11-03T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b8026ff420ed", "title": "TensorFlow . Probability : Hyperparameter Search", "url": "https://mdda.net/blog/research/talks/tensorflow-probability-hyperparameter-search", "published_at": "2019-10-10T00:00:00+00:00" }, { "id": "01a087a3-49df-70d7-9a69-b802703268ef", "title": "AI Day.TH : Transformers and Unicorns", "url": "https://mdda.net/blog/research/talks/ai-day-th-transformers-and-unicorns", "published_at": "2019-09-29T00:00:00+00:00" } ] posts Claim your blog

Back to mdda.net

Blog · corpus.blog/blogs/mdda.net/posts

mdda.net

mdda.net

2026

Harnesses for Generalisation original ↗

6 Aug 2026

GDG I/O Extended Singapore : New Gemma Models original ↗

27 Jun 2026

Local Models & Self-Evolving Harnesses original ↗

25 Jun 2026

DSTA BrainHack event - Invited Talk original ↗

10 Jun 2026

External Keypad for OBS Controls original ↗

28 May 2026

AI Engineer Gemini Event original ↗

13 May 2026

Running ~30B MoE models on an old GTX 1080 original ↗

11 May 2026

Robots : Reasoning and Acting original ↗

29 Apr 2026

Personal Agents - Cloud and Edge original ↗

26 Feb 2026

Signals Between the Tokens original ↗

24 Feb 2026

Learning to Think, Not Learning to Solve original ↗

17 Feb 2026

NeurIPS Roundup original ↗

15 Jan 2026

2025

Sizing gemma3 for TPUv5-n original ↗

16 Dec 2025

Practical TPU VM on GCP original ↗

15 Dec 2025

Reinforcement Learning for Long-Horizon Multi-Turn Search Agents original ↗

6 Dec 2025

Evolving Reasoning original ↗

26 Nov 2025

DevFest SG Talk: Lessons from Reinforcement Learning original ↗

22 Nov 2025

Neural Assets and World Models original ↗

15 Oct 2025

IPhO Gold using Agentic Gemini original ↗

25 Sept 2025

Agent Efficiency, Memory and Confidence original ↗

27 Aug 2025

A Reasoning-Based Approach to Cryptic Crossword Clue Solving original ↗

16 Jul 2025

GPU Kernel Scientist : An LLM-Driven Framework for Iterative Kernel Optimization original ↗

3 Jul 2025

Self-Improving Agents original ↗

19 Jun 2025

Evolving GPU Kernels original ↗

22 May 2025

Latent Space Reasoning original ↗

30 Apr 2025

Diffusion LLMs original ↗

25 Mar 2025

Getting to Aha (notes)original ↗

28 Feb 2025

Getting to Aha original ↗

25 Feb 2025

NeurIPS recap and SOTA original ↗

22 Jan 2025

2024

System 2 - Training for Test-time compute original ↗

26 Sept 2024

Slack to Mattermost on New Fedora server original ↗

20 Sept 2024

o1 - Are the curves awesome?original ↗

15 Sept 2024

Fedora 40 XFCE Live modifications original ↗

14 Sept 2024

Install cuda-12.6 on Fedora 40 original ↗

14 Sept 2024

Upgrade to Fedora 40 with cuda-12.4 original ↗

30 Aug 2024

Proving that Cryptic Crossword Clue Answers are Correct original ↗

27 Jul 2024

Small LM Agents original ↗

17 Jul 2024

Cryptic Proof Checking original ↗

18 Jun 2024

Upgrade to Fedora 38 with cuda-12.4 original ↗

10 May 2024

Long Context Transformers original ↗

25 Apr 2024

DSPy - Gemma Sprint original ↗

8 Apr 2024

Gemini 1.5 One Million Sprint original ↗

23 Mar 2024

DSPy - Gemini and Gemma original ↗

14 Mar 2024

Gemini 1 million original ↗

20 Feb 2024

Research from NeurIPS original ↗

18 Jan 2024

2023

Enable bluetooth management via GUI on Fedora original ↗

4 Dec 2023

DevFest SG Talk: No-code Prototyping with Flowise original ↗

18 Nov 2023

What's Next in RAG original ↗

16 Nov 2023

Converting a Presentation Transcript to Speaker Notes original ↗

27 Oct 2023

LLMs and Reasoning original ↗

14 Sept 2023

Keras Core original ↗

9 Sept 2023

Upgrade to Fedora 38 including Nvidia, etc original ↗

28 Jul 2023

Open-Endedness and Large Language Models original ↗

27 Jul 2023

Self-Improvement (for Language Models)original ↗

13 Jul 2023

Code and Reasoning with LLMs original ↗

26 Jun 2023

MediaPipe with a bit of Bard original ↗

28 May 2023

Images+LLMs : Advances in Multimodal Models original ↗

25 May 2023

Interacting with LLMs in a post-chatGPT world original ↗

20 Apr 2023

Cloud Platforms & Other Super Fun Stuff original ↗

13 Mar 2023

News You Can Use - Practical Stuff original ↗

23 Feb 2023

Voices In, Voices Out - ASR, TTS original ↗

19 Jan 2023

2022

Prompting, Instructions and the Future of Large Language Models original ↗

29 Nov 2022

World of Diffusion : Theory, Shortcuts & Future original ↗

15 Sept 2022

Running MineRL within a Container original ↗

13 Sept 2022

Containers within a GCP host server, with Nvidia pass-through original ↗

12 Sept 2022

GCP VM with Nvidia set-up and Convenience Scripts original ↗

2 Sept 2022

What’s New in AI & Machine Learning original ↗

18 Aug 2022

RETRO vs LaMDA - Clash of the Transformers original ↗

25 Jan 2022

Text Similarity for NLP original ↗

7 Jan 2022

2021

Back to Basics 03 original ↗

16 Dec 2021

Back to Basics 02 original ↗

9 Dec 2021

Back to Basics 01 original ↗

2 Dec 2021

GPT : Beams, Sampling & more original ↗

25 Nov 2021

Investigating Prompt Engineering in Diffusion Models original ↗

21 Nov 2021

Using a GCP VM like a 'local' Deep Learning machine original ↗

10 Oct 2021

Adding Firewalled Jupyter to a GCP VM original ↗

7 Oct 2021

Building a reusable Deep Learning VM on Google Cloud original ↗

5 Oct 2021

Nvidia (11.3) for TensorFlow & PyTorch while upgrading Fedora 34 original ↗

4 Sept 2021

PerceiverIO original ↗

26 Aug 2021

Exploring Different Approaches for Multilingual Tasks original ↗

5 Aug 2021

Back to Basics 03 original ↗

20 Jul 2021

Back to Basics 02 original ↗

13 Jul 2021

Back to Basics 01 original ↗

6 Jul 2021

Multi-Hop Inference Explanation Regeneration by Matching Expert Ratings original ↗

11 Jun 2021

Positioning Yourself for the Future: AI from the Backend original ↗

10 Apr 2021

TensorFlow Everywhere original ↗

18 Feb 2021

2020

LIT : LSTM-Interleaved Transformer for Multi-Hop Explanation Ranking original ↗

13 Dec 2020

Where the jobs are in AI, and Why original ↗

12 Sept 2020

TensorFlow 2.3 from source on Fedora 32 original ↗

29 Jul 2020

Deep Learning State of the Art original ↗

6 Feb 2020

2019

TensorBoard Lite original ↗

19 Dec 2019

TensorFlow 2.1 from source on Fedora 30 original ↗

4 Dec 2019

Faces in the Browser using TensorFlow.js original ↗

30 Nov 2019

Neural Structured Learning original ↗

12 Nov 2019

Nvidia through docker on Fedora 30 original ↗

12 Nov 2019

Language Model Assisted Explanation Generation original ↗

4 Nov 2019

Paraphrasing with Large Language Models original ↗

4 Nov 2019

Unsupervised Natural Question Answering with a Small Model original ↗

3 Nov 2019

TensorFlow . Probability : Hyperparameter Search original ↗

10 Oct 2019

AI Day.TH : Transformers and Unicorns original ↗

29 Sept 2019