corpus.blog
Most cited
Talked about
Blogs
41554 blogs · [ { "id": "01a08d43-f585-73e5-8974-f9fd268b5029", "title": "Extending Raschka's GPT-2: an MoE trained from scratch on an RTX 3090", "url": "https://www.gilesthomas.com/2026/09/gpt-2-to-moe", "published_at": "2026-09-10T18:45:00+00:00" }, { "id": "01a0877d-0b30-706b-95b7-1bd03dffacbd", "title": "Putting my JAX-trained models on the Hugging Face Hub", "url": "https://www.gilesthomas.com/2026/09/jax-models-on-hugging-face", "published_at": "2026-09-03T18:00:00+00:00" }, { "id": "01a0877d-0b30-706b-95b7-1bd03d38a225", "title": "Why do OpenAI's GPT-2 weights beat mine? Part four: digging into dropout", "url": "https://www.gilesthomas.com/2026/08/why-do-openai-gpt2-weights-beat-mine-4-ift-dropout", "published_at": "2026-08-27T19:00:00+00:00" }, { "id": "01a0877d-0b30-706b-95b7-1bd03caaba53", "title": "Adding diagrams to my static site generator with D2", "url": "https://www.gilesthomas.com/2026/08/adding-d2", "published_at": "2026-08-25T23:00:00+00:00" }, { "id": "01a0877d-0b30-706b-95b7-1bd03ca15462", "title": "Use the built-in GELU, don't roll your own!", "url": "https://www.gilesthomas.com/2026/08/built-in-gelu", "published_at": "2026-08-20T02:00:00+00:00" }, { "id": "01a0877d-0b30-706b-95b7-1bd03c0b72e8", "title": "A quick(ish) Chinchilla check", "url": "https://www.gilesthomas.com/2026/08/chinchilla-check", "published_at": "2026-08-07T19:00:00+00:00" }, { "id": "01a0877d-0b30-706b-95b7-1bd03b7eb09f", "title": "How I use AI on this blog", "url": "https://www.gilesthomas.com/2026/07/ai-use", "published_at": "2026-07-31T19:00:00+00:00" }, { "id": "01a0877d-0b30-706b-95b7-1bd03b0bdc70", "title": "Why do OpenAI's GPT-2 weights beat mine? Part three: testing overtraining", "url": "https://www.gilesthomas.com/2026/07/why-do-openai-gpt2-weights-beat-mine-3-overtraining", "published_at": "2026-07-31T01:15:00+00:00" }, { "id": "01a0877d-0b30-706b-95b7-1bd03a945030", "title": "Why do OpenAI's GPT-2 weights beat mine? Part two: the bugfix", "url": "https://www.gilesthomas.com/2026/07/why-do-openai-gpt2-weights-beat-mine-2-the-bugfix", "published_at": "2026-07-30T19:00:00+00:00" }, { "id": "01a0877d-0b30-706b-95b7-1bd03a1ef9b9", "title": "Why do OpenAI's GPT-2 weights beat mine?", "url": "https://www.gilesthomas.com/2026/07/why-do-openai-gpt2-weights-beat-mine-1-intro", "published_at": "2026-07-29T15:30:00+00:00" }, { "id": "01a0877d-0b2f-70d1-a88c-a0ab1c89c7f1", "title": "Benchmarking Qwen 3.6 35B MoE (3B active) on an RTX 3090", "url": "https://www.gilesthomas.com/2026/07/benchmarking-qwen-3-6-35b-moe-rtx-3090", "published_at": "2026-07-24T23:30:00+00:00" } ] posts
Claim your blog
Back to Giles' blog
Blog · corpus.blog/blogs/gilesthomas.com/posts
Giles' blog
gilesthomas.com
2026
Extending Raschka's GPT-2: an MoE trained from scratch on an RTX 3090
original ↗
10 Sept 2026
Putting my JAX-trained models on the Hugging Face Hub
original ↗
3 Sept 2026
Why do OpenAI's GPT-2 weights beat mine? Part four: digging into dropout
original ↗
27 Aug 2026
Adding diagrams to my static site generator with D2
original ↗
25 Aug 2026
Use the built-in GELU, don't roll your own!
original ↗
20 Aug 2026
A quick(ish) Chinchilla check
original ↗
7 Aug 2026
How I use AI on this blog
original ↗
31 Jul 2026
Why do OpenAI's GPT-2 weights beat mine? Part three: testing overtraining
original ↗
31 Jul 2026
Why do OpenAI's GPT-2 weights beat mine? Part two: the bugfix
original ↗
30 Jul 2026
Why do OpenAI's GPT-2 weights beat mine?
original ↗
29 Jul 2026
Benchmarking Qwen 3.6 35B MoE (3B active) on an RTX 3090
original ↗
24 Jul 2026