6). Logic-of-Thought – proposes a new prompting technique called Logic-of-Thought (LoT) which employs propositional logic to generate and inject expanded logical information from the input context.
RESEARCH
-
AlphaChip: Reinforcement Learning Designs Google TPU Physical Layout
By
–
3). AlphaChip – a reinforcement learning-based method trained to design the physical layout of chips; AlphaChip is reportedly used in three additional generations of Google’s TPU…https://t.co/kM6HcgSTj4
— DAIR.AI (@dair_ai) 29 septembre 20243). AlphaChip – a reinforcement learning-based method trained to design the physical layout of chips; AlphaChip is reportedly used in three additional generations of Google’s TPU…
-
Large Language Models Struggle with Domain-Independent Planning Tasks
By
–
4). LLMs Still Can’t Plan – evaluates whether large reasoning models such as o1 can plan; finds that a domain-independent planner can solve all instances of Mystery Blocksworld but LLMs struggle, even on small instances…
-
SolarLLM Podcast: Google AI’s Latest Language Model
By
–
PodCast about #SolarLLM by @GoogleAI #NotebookLM.
— Sung Kim (@hunkims) 29 septembre 2024
Also check out https://t.co/soZOfwlhzb. pic.twitter.com/ggO1uO7JKgPodCast about #SolarLLM by @GoogleAI #NotebookLM. Also check out https://
huggingface.co/upstage/solar-
pro-preview-instruct
…. -
Claude vs Gemini: Choosing the Right Language Model
By
–
Claude vs Gemini: A Guide to Choose the Right Model: Introduction Within the quickly changing field of artificial intelligence, two language models, Claude and Gemini,… https://
analyticsvidhya.com/blog/2024/09/c
laude-vs-gemini/?utm_source=dlvr.it&utm_medium=twitter
… #DataAnalytics #DataScience #DataDriven #CIO #BigData #AI #CTO #ArtificialIntelligence -
Model Merging Techniques and Hidden Representation Dynamics
By
–
I'd say it barely works tbh… 🙂 It really depends on the type of merge and whether you retrain it or not. In this case, it's not really "working". The additional layers seem to push the hidden representations into new areas, which may or may not be interesting, depending on
-
Distillation vs DPO: Model Optimization Techniques Comparison
By
–
Yes, for sure. Distillation would be ideal but DPO would also work.
-
Google Showcases 70+ Papers at ECCV 2024 Milan
By
–
This year Milan, Italy hosts #ECCV2024, which kicks off today with a full docket of workshops and tutorials. Google has a strong presence at @eccvconf with over 70 accepted papers and participation in over 25 workshops and tutorials. Find the full list @ https://
goo.gle/eccv24 -
Model Merges and 34B Evaluation Updates
By
–
I would usually assume that but my merges do work though. It might have been fixed after the 34B eval?
-
Open Multimodal LLM Releases Close Gap with Closed-Source Models
By
–
Celebrating a crazy month of "Open Multimodal LLM Releases"! If you haven't followed it several research labs have release impressively capable open multimodal LLMs in September 2024: Qwen-VL, Pixtral, LLama 3.2 and Molmo, seriously closing the gap with closed-source model