Tx_LLM_A_Large_Language_Model_for_Therapeutics_1720524147.pdf
1.1 MB
Promising news from #Google in the domain of generative pharma
“Developing #therapeutics is a lengthy and expensive process that requires the satisfaction of many different criteria, and #AI models capable of expediting the process would be invaluable.
However, the majority of current AI approaches address only a narrowly defined set of tasks, often circumscribed within a particular domain.
To bridge this gap, we introduce Tx-LLM, a generalist LLM fine-tuned from PaLM-2 which encodes knowledge about diverse therapeutic modalities."
#pharma #nextpharma #healthcare #artificialintelligence #google
“Developing #therapeutics is a lengthy and expensive process that requires the satisfaction of many different criteria, and #AI models capable of expediting the process would be invaluable.
However, the majority of current AI approaches address only a narrowly defined set of tasks, often circumscribed within a particular domain.
To bridge this gap, we introduce Tx-LLM, a generalist LLM fine-tuned from PaLM-2 which encodes knowledge about diverse therapeutic modalities."
#pharma #nextpharma #healthcare #artificialintelligence #google
AI molecular phenotyping has a new business model. Tempus AI received the first-ever CPT code for an algorithm-only analysis of previously-sequenced, laboratory developed test trannoscriptomic data.
The PurISTSM test uses RNA profiling to identify molecular subtypes in patients with advanced pancreatic ductal adenocarcinoma, which may inform first-line therapy selection and management.
It will be interesting to see how payers assess the "medical necessity" and therefore reimbursability of clinical AI as these models come to market.
The PurISTSM test uses RNA profiling to identify molecular subtypes in patients with advanced pancreatic ductal adenocarcinoma, which may inform first-line therapy selection and management.
It will be interesting to see how payers assess the "medical necessity" and therefore reimbursability of clinical AI as these models come to market.
Tempus Labs, Inc.
American Medical Association Grants PLA Code to Tempus Algorithmic Test, PurISTSM | Tempus Labs, Inc.
CHICAGO --(BUSINESS WIRE)--Jul. 8, 2024-- Tempus AI, Inc. (NASDAQ: TEM), a leader in artificial intelligence and precision medicine, today announced that the company’s PurIST SM algorithmic test has received a proprietary laboratory analysis (PLA) code from…
#China plans to boost national compute capacity 30% (from 230 Exaflops to 300 EF) by 2025
China currently has over 8.1 million datacenter racks in operation.
China currently has over 8.1 million datacenter racks in operation.
The Register
China plans to boost national compute capacity thirty percent by 2025
From 230 Exaflops to 300, with Tesla a part of the plan for energy storage, - and cars
Anthropic announced Claude's Artifacts can be published online.
This just changed online education forever.
Every single piece of content, screenshot, PDF, presentation, etc., can now be turned into an interactive learning game.
Here's a basic example:
1. Imported an entire 50-minute workshop we did at The Rundown AI University on creating commercials using AI (Trannoscript + the PowerPoint presentation)
2. Gave Claude Sonnet 3.5 this prompt:
"Create an interactive learning game web app in the form of a quiz based on the PDF and Trannoscript of a workshop attached. After each question, provide an explanation of why the answer is correct, citing relevant parts of the original text."
3. Let Claude code and generate the interactive game for you (it took ~1 minute for me)
4. Click Publish at the bottom right, and share the Artifact with your students so they can learn interactively
The number of use cases that just unlocked is wild. Imagine doing this at scale for practicing for exams, interview screening, corporate training, employee onboarding, etc.
This just changed online education forever.
Every single piece of content, screenshot, PDF, presentation, etc., can now be turned into an interactive learning game.
Here's a basic example:
1. Imported an entire 50-minute workshop we did at The Rundown AI University on creating commercials using AI (Trannoscript + the PowerPoint presentation)
2. Gave Claude Sonnet 3.5 this prompt:
"Create an interactive learning game web app in the form of a quiz based on the PDF and Trannoscript of a workshop attached. After each question, provide an explanation of why the answer is correct, citing relevant parts of the original text."
3. Let Claude code and generate the interactive game for you (it took ~1 minute for me)
4. Click Publish at the bottom right, and share the Artifact with your students so they can learn interactively
The number of use cases that just unlocked is wild. Imagine doing this at scale for practicing for exams, interview screening, corporate training, employee onboarding, etc.
❤2
Modular computation in neural networks
Researchers 've explored how artificial recurrent networks handle multiple tasks, offering insights into flexible computation.
Key finding: "dynamical motifs" - recurring patterns of neural activity that implement specific computations. These motifs, like attractors and decision boundaries, are reused across different tasks.
This work proposes dynamically motifs as a fundamental unit of compositional computation, bridging the gap between individual neurons and whole networks.
It offers a new framework for understanding specialization and generalization in neural systems.
Researchers 've explored how artificial recurrent networks handle multiple tasks, offering insights into flexible computation.
Key finding: "dynamical motifs" - recurring patterns of neural activity that implement specific computations. These motifs, like attractors and decision boundaries, are reused across different tasks.
This work proposes dynamically motifs as a fundamental unit of compositional computation, bridging the gap between individual neurons and whole networks.
It offers a new framework for understanding specialization and generalization in neural systems.
Nature
Flexible multitask computation in recurrent networks utilizes shared dynamical motifs
Nature Neuroscience - The authors identify reusable ‘dynamical motifs’ in artificial neural networks. These motifs enable flexible recombination of previously learned capabilities,...
Huge synthetic dataset from Tencent AI Lab.
This looks awesome to fiddle with. Code to synthesize with GPT-4o + VLLM
HuggingFace
GitHub
This looks awesome to fiddle with. Code to synthesize with GPT-4o + VLLM
HuggingFace
GitHub
huggingface.co
proj-persona/PersonaHub · Datasets at Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
AI_Revolution_in_Education_2024_1720624353.pdf
7.4 MB
The World Bank analyzed the impact of #AI in #education
The AI revolution is rapidly transforming the sector, offering personalized learning, teacher support, and optimized management.
Recent research shows AI, like GPT-4, significantly boosts #productivity in knowledge work.
AI's rapid advancement now enables complex conversations, human-like text generation, and multimodal content creation.
The AI revolution is rapidly transforming the sector, offering personalized learning, teacher support, and optimized management.
Recent research shows AI, like GPT-4, significantly boosts #productivity in knowledge work.
AI's rapid advancement now enables complex conversations, human-like text generation, and multimodal content creation.
👍1
AI chip developer Groq is nearing a deal with BlackRock that could give it a $2.2Bn pre-money valuation.
It's a good sign for Nvidia rivals, but BlackRock is making Groq submit to a financial audit for the deal.
More information about Groq is here.
It's a good sign for Nvidia rivals, but BlackRock is making Groq submit to a financial audit for the deal.
More information about Groq is here.
The Information
Groq, an Nvidia Rival, Nears $2.2 Billion-Valuation BlackRock Deal; Musk’s Mystery Company Revealed
The startups developing specialized artificial intelligence chips face some heady challenges—most obviously that they’re taking on entrenched giant Nvidia. One challenge they don’t face, though, is a lack of funding.Groq, a startup that develops chips that…
Elon Musk and Neuralink went live and shared the insane potential of Neuralink <> Tesla Optimus
1. People who can't speak can communicate with Optimus
2. People who've lost limbs could possibly use a robotic Optimus limb and control it with their brains
1. People who can't speak can communicate with Optimus
2. People who've lost limbs could possibly use a robotic Optimus limb and control it with their brains
The winning AI Math Olypiad model is out!
NuminaMath-7B-TIR, the small but mighty model that won the first progress prize of the AI Math Olympiad 🥇!
> Fine-tuned with iterative SFT on DeepSeekMath-7B from
> Stage 1: learn math with chain of thought samples
> Stage 2: learn code with tool-integrated reasoning (TIR)
> Inference: self-consistency decoding with tool-integrated reasoning to generate solutions.
Demo.
NuminaMath-7B-TIR, the small but mighty model that won the first progress prize of the AI Math Olympiad 🥇!
> Fine-tuned with iterative SFT on DeepSeekMath-7B from
> Stage 1: learn math with chain of thought samples
> Stage 2: learn code with tool-integrated reasoning (TIR)
> Inference: self-consistency decoding with tool-integrated reasoning to generate solutions.
Demo.
huggingface.co
AI-MO/NuminaMath-7B-TIR · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
Exscientia launched a drug discovery platform in partnership with AWS.
The platform combines both design of drug candidates and automation that can speed up molecular synthesis
The platform combines both design of drug candidates and automation that can speed up molecular synthesis
investors.exscientia.ai
Exscientia Launches AWS AI-powered Platform to Advance Drug Discovery
— State-of-the-art platform, built using Amazon Web Services technologies, integrates generative AI drug design and robotic lab automation to further accelerate Exscientia’s ability to deliver high quality drug candidates at faster speed and lower cost —…
Bioptimus released of H-optimus-0, the world’s largest open-source AI foundation model for pathology that achieves SOTA performances in many important downstream tasks.
GitHub
releases/models/h-optimus at main · bioptimus/releases
Contribute to bioptimus/releases development by creating an account on GitHub.
Law_Commission_DAOs_Scoping_Paper_1720711449.pdf
6.2 MB
Decentralized autonomous organizations (DAOs). A scoping paper
This report explores benefits that could accrue from the use of the LEI in cross-border payment systems and, in close coordination with the Global LEI Foundation, the Regulatory Oversight Committee and national authorities, explores options to promote broader LEI adoption.
It builds on a survey among FSB authorities and an outreach event with market participants, both held in the first half of 2022.
It also extensively benefitted from inputs provided by GLEIF and ROC, including the ROC’s 2018 and 2019-2021 Progress Reports.
This report explores benefits that could accrue from the use of the LEI in cross-border payment systems and, in close coordination with the Global LEI Foundation, the Regulatory Oversight Committee and national authorities, explores options to promote broader LEI adoption.
It builds on a survey among FSB authorities and an outreach event with market participants, both held in the first half of 2022.
It also extensively benefitted from inputs provided by GLEIF and ROC, including the ROC’s 2018 and 2019-2021 Progress Reports.
🔥2
A neuromodulation market analysis from LSI's Market Intelligence
LSI's Market Intelligence team just released an article on the Neuromodulation market, valuing it at ~$6.6 billion in 2023 and projecting a 7.7% CAGR from 2023 to 2028.
Drivers of this include growth in devices and emerging companies to treat a breadth of disease states:
Incontinence
• Axonics, Inc. (acq. Boston Scientific), and their therapy to treat overactive bladder and fecal incontinence
• Amber Therapeutics, and their implantable therapy to address stress and treating urinary incontinence and overactive bladder
Depression
• Sooma Medical, and their transcranial direct current stimulation (tDCS) device for at-home treatment of depression
• Ampa Medical, and their NextGen TMS™ (transcranial magnetic stimulation) device to treat people with medication-resistant depression
• Inner Cosmos, and their implantable neuromodulation devices (promising findings from initial phase of its human trials in April 2024)
Alzheimer's
• Cognito Therapeutics, Inc., and their therapy designed to evoke gamma oscillation with non-invasive visual and auditory stimulation
• Sinaptica Therapeutics, and their personalized closed-loop neuromodulation therapy to enhance neuroplasticity through stimulation of key brain networks involved in memory (promising Phase 2 placebo-controlled results slowing disease progression by 80%+ at 6 months)
LSI's Market Intelligence team just released an article on the Neuromodulation market, valuing it at ~$6.6 billion in 2023 and projecting a 7.7% CAGR from 2023 to 2028.
Drivers of this include growth in devices and emerging companies to treat a breadth of disease states:
Incontinence
• Axonics, Inc. (acq. Boston Scientific), and their therapy to treat overactive bladder and fecal incontinence
• Amber Therapeutics, and their implantable therapy to address stress and treating urinary incontinence and overactive bladder
Depression
• Sooma Medical, and their transcranial direct current stimulation (tDCS) device for at-home treatment of depression
• Ampa Medical, and their NextGen TMS™ (transcranial magnetic stimulation) device to treat people with medication-resistant depression
• Inner Cosmos, and their implantable neuromodulation devices (promising findings from initial phase of its human trials in April 2024)
Alzheimer's
• Cognito Therapeutics, Inc., and their therapy designed to evoke gamma oscillation with non-invasive visual and auditory stimulation
• Sinaptica Therapeutics, and their personalized closed-loop neuromodulation therapy to enhance neuroplasticity through stimulation of key brain networks involved in memory (promising Phase 2 placebo-controlled results slowing disease progression by 80%+ at 6 months)
OpenAI has come up w/ a framework of 5 levels to track progress twd AGI, and think they're currently near level 2 ("Reasoners")
At recent all-hands, leadership also did a research demo of GPT4 w/ improved reasoning
At recent all-hands, leadership also did a research demo of GPT4 w/ improved reasoning
👍4⚡3
Graphcore has been acquired by SoftBank. Masayoshi Son laid out his grand vision for artificial superintelligence last month, and it now seems clear Graphcore will be a part of that plan.
EE Times
AI Chip Startup Graphcore Acquired by SoftBank
The British chipmaker will join SoftBank as part of chairman and CEO Masayoshi Son’s grand vision for artificial superintelligence.
Patronus AI announced the release of 'Lynx'. It's a new open-source hallucination detection model.
- Beats GPT-4o on hallucination tasks
- Open source, open weights, open data
- Excels in real-world domains like medicine and finance.
The startup claims it outperforms existing solutions such as GPT-4, Claude-3-Sonnet, and other models used as judges in closed and open-source settings.
Hf.
Paper.
- Beats GPT-4o on hallucination tasks
- Open source, open weights, open data
- Excels in real-world domains like medicine and finance.
The startup claims it outperforms existing solutions such as GPT-4, Claude-3-Sonnet, and other models used as judges in closed and open-source settings.
Hf.
Paper.
www.nomic.ai
Evaluating LLM Hallucination Benchmarks with Embeddings
⚡2
A really nice read on the wider set of mathematical tools and approaches that can help model data with perhaps less standard topological or algebraic properties
arXiv.org
Beyond Euclid: An Illustrated Guide to Modern Machine Learning...
The enduring legacy of Euclidean geometry underpins classical machine learning, which, for decades, has been primarily developed for data lying in Euclidean space. Yet, modern machine learning...
Commure announced Strongline Copilot - a wearable AI device for nurses, physicians, and healthcare administrators to interact with their EMR, automatically generate SOAP notes, and save valuable time in-clinic.
It's compatible with Strongline platform.
It's compatible with Strongline platform.
Commure
Scribe Co-Pilot | Commure
Discover the leading AI device for medical documentation, revolutionizing healthcare with seamless, accurate, and efficient record-keeping.