#codemixeddata hasil pencarian
Congratulations @ShrimalAnubhav and Siddharth Yadav on getting papers in #AAAI21 Student Abstract and Poster Program (SA-21), to be held during @RealAAAI. @IIITDelhi #codemixeddata #imagecaptioning #unsupervisedlearning #MachineLearning #DeepLearning
Researchers squeezed GPT-2-class performance out of a model trained on just 1 billion tokens - 10× less data - by dialing in a sharp dataset mix: 50% finePDFs, 30% DCLM-baseline, 20% FineWeb-Edu... huggingface.co/blog/codelion/… --- Want similar stories? Join 👉 faun.dev/join
yeah i tried it for a few tasks and it's pretty bad (small sample size). the only reason to use codex for coding was because it was super smart, i don't see a point using one that's faster but dumber since other models already do that with better tooling around them (e.g. claude…
Code merging: where two branches meet for coffee and end up in a bar fight. You start optimistic, end up negotiating peace treaties between curly braces, and question life choices while CI/CD plays judge. #DevLife #CodeMergeChaos
🚨 The Contradiction Conundrum: How Mixed Security Messages Are Creating Your Next #Data Breach undercodetesting.com/the-contradict… Educational Purposes!
check out this amazing data mixer approach by @MayeeChen
Thrilled to have contributed to Olmo 3! The best fully open 32B model (data, training recipes, checkpoints and more!) As an intern at AI2 these last 8 months, I’ve grown to deeply appreciate the careful science, iteration, and collaboration that go into models like this and have…
OpenAI continues their tradition of "great bar charts". In the Codex usage dashboard this chart has no Y-axis. No clue what these bars mean at all other than a vague idea of usage relative to other days.
🍣Data mixing is a little too powerful It's really easy to accidentally learn "optimal" mixes that oversample from certain pockets heavily. For example, "STEM" docs are really valuable for climbing tasks like MMLU & mixing methods can propose very high weight. But…
🔬 Linear Mixed Effects Models Repeated measures in clinical trials = correlated data points. Are you using the right model? 🔍 Learn how linear mixed effects models help analyse repeated measures data 👇: fiosgenomics.com/linear-mixed-e… #bioinformatics #clinicaldata #datascience
Every year, ER visits spike after Thanksgiving celebrations due to kitchen mishaps, overindulgence, and wintry weather. Code’s CortexDecoder SDK helps clinicians process patient data quickly and accurately to document and deliver quality care. Here’s how:…
been pondering how that mix could sharpen the agent handoffs in my codemachine cli experiments. github.com/moazbuilds/Cod…
Your provided series diverges from CODATA's 137.035999084(21) starting at the 10th decimal—206 vs. 084—failing basic accuracy beyond perturbative QED's reach. True theoretical superiority demands matching measured values to arbitrary precision without post-hoc adjustment,…
What's wild is the training data mix. Web crawls, code repos, media, user interactions, commercial datasets, AND synthetic AI-generated content. This is how you build a model that understands actual workflows, not just academic benchmarks.
Another codemode experiment that seems to improve quality somewhat: For tools that don't have helpful output types (either internal, like many of ours, or from MCP servers like Linear's), generate types based on actual outputs from prior calls.
Day 8 @CodeAutomation @Google shares TUMIX, a smart way for AI agents to team up. They run in parallel, one codes, one searches, then chat to refine the best answer. Smarter results with less power used. Agents collaborating like pros! #GoogleTUMIX #AIAgents #CodeAutomation
Basically, analyzing any sets of histograms (normalized) using this underlying geometry. Related to Compositional Data Analysis (CoDA)
初めてCodexを使って、授業の受講生同士のネットワーク調査データをクリーニングして可視化してもらった。何度かやり取りしたが、AIが自動でRでコーディングして、とても時短になる。 まずは小分けできるちょっとした作業で使いつつ、段々と自分の研究プロジェクトでもフルに使いたい。
Rohan, that's a brilliant observation! It seems like sometimes the messier the data, the better the final result, right?
Congratulations @ShrimalAnubhav and Siddharth Yadav on getting papers in #AAAI21 Student Abstract and Poster Program (SA-21), to be held during @RealAAAI. @IIITDelhi #codemixeddata #imagecaptioning #unsupervisedlearning #MachineLearning #DeepLearning
Something went wrong.
Something went wrong.
United States Trends
- 1. #AEWFullGear 10.2K posts
- 2. #RiyadhSeason 15.3K posts
- 3. Georgia Tech 3,828 posts
- 4. Mason 39.1K posts
- 5. Utah 18.7K posts
- 6. Syracuse 8,691 posts
- 7. Kansas State 4,015 posts
- 8. Bam Rodriguez 3,130 posts
- 9. Okada 9,831 posts
- 10. #AEWTailgateBrawl 3,020 posts
- 11. #TheRingIV 5,517 posts
- 12. Lincoln Riley 1,751 posts
- 13. Haney 6,634 posts
- 14. Oregon 28.6K posts
- 15. Utes 1,516 posts
- 16. Avery Johnson N/A
- 17. Ethan Davis N/A
- 18. Arch Manning 4,109 posts
- 19. #Boxing 5,741 posts
- 20. Joe Jackson 1,234 posts