David Talby @davidtalby

PhD, MBA, CTO @JohnSnowLabs. Making #NLP, #AI, and #DataScience solve real-world problems in #Healthcare & #Pharma. Views are my own. linkedin.com/in/davidtalby Seattle, WA Joined June 2009

Tweets

3K
Followers

27K
Following

11K
Likes

142K

David Talby @davidtalby

2 years ago

These awards are based on interviews with our customers. Thank you for the recognition and support! lnkd.in/gtzpC5Mk

0 2 11 1K 0

David Talby @davidtalby

2 years ago

Defining 3 criteria for Regulatory-Grade #AI: Truthfulness ("No BS"), Responsible AI, and Private Execution. #responsibleai #regulation #ethicalai #datascience lnkd.in/gc3MErSa

0 2 2 1K 0

#SparkNLP 4.4.1 is out! New DistilBertForZeroShotClassification annotator: List which classes you want text to be classified by, and just process documents - no training data or model tuning is required. 100% Free & #OpenSource. Getting started:…lnkd.in/e8fXu4TX

0 1 3 1K 0

David Talby @davidtalby

2 years ago

Compressing #LargeLanguageModels Prompts with Gist Tokens: "Finetuning and distillation methods allow for specialization of LMs without prompting, but require retraining the model for each task. To avoid this trade-off entirely, we present gisting, whic…lnkd.in/gE5EHkQk

0 2 1 776 1

David Talby @davidtalby

2 years ago

This Tuesday (April 25th, noon EDT) at a webinar of the #AMIA #NLP working group, I'm presenting ‘2023 Trends in Applied NLP in Healthcare: Large Language Models, No-Code, and Responsible AI’ Free registration: #Healthcare #Heal…lnkd.in/gTGAX_ZU lnkd.in/gvFUXcJn

0 1 2 751 1

David Talby @davidtalby

2 years ago

Databricks article by @MaziyarPanahi on scaling up Vision Transformers. #SparkNLP shows near-linear speedup of 9.6x when scaling to 10 nodes - without changing a single line of code. 100% Free & #OpenSource. #ai #datascience #comp…lnkd.in/geuzm9N2 lnkd.in/gsHuV8ip

0 1 3 670 0

David Talby @davidtalby

2 years ago

#Legal #NLP 1.11 is out! Legal contract Summarization, 15 new entity recognition models for Nondisclosure Agreements, 5 Legal Clasification models, and Visual NLP demos for information extraction from scanned legal forms. lnkd.in/gBa7igfv #NLProc…lnkd.in/gkYPyfgj

0 1 3 808 1

David Talby @davidtalby

2 years ago

On Monday at the #HIMSS conference, Vishakha Sharma, PhD from @Roche will present at the #ML & #AI about using #NLP to enable personalized care in #digitalhealth products. lnkd.in/gH5VTEQz

0 0 3 704 0

David Talby @davidtalby

2 years ago

A Comparison of #SparkNLP for #Healthcare and #ChatGPT on Clinical Named Entity Recognition #ai #datascience #llms #largelanguagemodels #deeplearning #Healthcareai #healthai #textmining lnkd.in/g6-8m2vx

0 1 4 667 0

David Talby @davidtalby

2 years ago

Major release: #SparkNLP for #Healthcare 4.4 is out! Includes the new BioGPT-JSL #LLM and a suite of instruction-tuned language models for Medical Text Generation, Question Answering, and Summarization. Benchmarks suggest that healthcare-specific LLM's a…lnkd.in/gm527MwM

0 1 5 842 1

David Talby @davidtalby

2 years ago

#SparkNLP 4.4 it out! Major release adding BART for Text Translation & Summarization, ConvNeXT Transformer for Image Classification, Zero-Shot Text Classification, and 400+ state-of-the-art models. 100% Free & #OpenSource. * ConvNextForImageClassificatio…lnkd.in/gxjkXCX6

0 1 2 935 0

David Talby @davidtalby

2 years ago

"Cerner Enviza, an Oracle company, along with John Snow Labs, are now helping support the U.S. Food and Drug Administration’s drug safety Sentinel Initiative. By developing artificial intelligence (#AI) tools aimed at extracting…lnkd.in/gGu7YqKg lnkd.in/gSMac2rT

0 0 1 750 0

David Talby @davidtalby

2 years ago

#LowCode Versus #NoCode #AI: The Differences and Why They Matter lnkd.in/ggNmKTFp

1 2 4 848 1

David Talby @davidtalby

2 years ago

Introducing the #OpenSource Python library for testing #NLP models. Automatically generate tests, run tests, and augment data to make #ResponsibleAI a reality and deliver safe & effective models. 50+ out-of-the-box test types inclu…lnkd.in/grnYxzkk lnkd.in/gKQ8aKiq

0 3 5 1K 1

David Talby @davidtalby

2 years ago

The #Healthcare #nLPSummit starts tomorrow! 30+ sessions covering the latest medical large language models, responsible AI libraries, best practices and case studies from teams putting this technology to use. #ai #datascience #n…lnkd.in/gR6zMs-4 nlpsummit.org

1 1 6 376 0

David Talby @davidtalby

2 years ago

Nice write-up from Stanford's human-centered #AI group on the last month's trends in #LLMs: deployments on the rise, worsening transparency, massive influx of funding, and demands on policy & regulation. lnkd.in/grME6Hph lnkd.in/g7TWmykZ

0 1 2 640 0

David Talby @davidtalby

2 years ago

#Legal #NLP 1.10 is out! The new version adds EurVoc Taxonomy Classification; Legal Criticality & Argument Mining in English & German; and new Legal Question Answering demos; and Multilingual Legal Sentence and Word Embeddings for Italian, Portuguese, Spa…lnkd.in/gm9XTNUF

0 1 0 972 0

David Talby @davidtalby

2 years ago

#ChatGPT is more accurate than crowd-sourced human labeling (via mechanical turk) on a sample of 2,382 tweets for tasks including relevance, stance, topics, and frames detection. ChatGPT is also about 20 times cheaper than MTurk for these tasks. Paper:…lnkd.in/graG7P-W

0 0 3 529 0

David Talby @davidtalby

2 years ago

"We introduce Vicuna-13B, an open-source chatbot trained by fine-tuning LLaMA on user-shared conversations collected from ShareGPT. Preliminary evaluation using GPT-4 as a judge shows Vicuna-13B achieves more than 90%* quality of O…lnkd.in/gfMVubA4 vicuna.lmsys.org

0 1 4 457 0

David Talby @davidtalby

2 years ago

The #Healthcare #NLPSummit starts on Tuesday! 30+ technical sessions with a focus on medical #LargeLanguageModels including keynotes from Microsoft (BioGPT), Google (EHRSafe), Stanford CRFM (PubMedGPT), MIT, U Pittsburg, and John S…lnkd.in/gQcRdqQX nlpsummit.org