A reminder that even the best experts on LLMs don’t really understand all their behaviors (in this case, why AIs often seem a bit preachy & condescending). Lots left to learn.
LLMS
-
LangSmith Expands Custom Model Cost Tracking Capabilities
By
–
Custom Model Costs in LangSmith ✎ We've expanded token-based cost tracking to be fully customizable! This means that, in addition to tracking costs for OpenAI models, you can track costs for *any* model that you use. To register a model for cost tracking, navigate to
-
Maximizing Existing Models Over Chasing New Arrivals
By
–
A significant oversight I often see is the overemphasis on the arrival of new models, while underestimating the transformative possibilities that lie in creatively fusing together the models we already have
-
GenAI Builder Launch Event: Generative AI Impact on Enterprise Workflows
By
–
Watch now on-demand the recording of the @SnapLogic GenAI Builder launch event: https://
snaplogic.com/resources/webc
asts/introducing-genai-builder?utm_source=TW&utm_medium=SOC&utm_campaign=2024_0117_ONL_CORP_GenAIBuilder-Webinar
…
…Other panelists and I discuss the impacts of #GenerativeAI and #LLMs on workflows, processes, products & services for multiple enterprise use cases and lines of business. -
Understanding Consistent Behavior Across Independent Language Models
By
–
I’d love to understand this better too… I thought it was just a quirk of the specifics of labeling instructions, but then multiple (what I think should be mostly independent) language models seem to all do this.
-
Measuring LLM Abilities: Current Limitations and Testing Challenges
By
–
We don't know how to measure LLM abilities well. Most tests are groups of multiple choice questions, tasks, or trivia – they don't represent real world uses well, they are subject to gaming & results are impacted by prompt design in unknown ways. Or they use human preference.
-
Enterprise LLM Adoption Surges: 38% to 7% Zero-Production Gap
By
–
38% of respondents said their organization had zero production applications backed by customized LLMs in 2023. 7% said they expected that to be the case by the end of 2024. That's a big leap! See more in our survey results. #genai #llm #enterpriseai
-
Building JSON Agents with Ollama LangChain Neo4j
By
–
JSON-based Agents With Ollama & LangChain Learn to implement a Mixtral agent that interacts with a graph database Neo4j through a semantic layer This work by @tb_tomaz is great for a few reasons: – Shows how to build an agent with an OSS model
– Shows how to build and use a -
GPT-Fast Integrates Gemma with Optimized Token Performance
By
–
+/-12 lines of code in gpt-fast integrated Gemma from @GoogleDeepMind ; and shows:
* 234 tokens / sec on V100 in int8
* 144 tokens / sec on V100 in float precision https://
github.com/pytorch-labs/g
pt-fast/commit/ef055fc12188eaf80d8ba948ad743ee5583d0f3c
… -
Semantic Chunking: Three Methods for Document Splitting
By
–
Semantic Chunking Semantic chunking using embedding distances between sentences to figure out how to split a document. A key decision is what threshold to split at. We now support three methods: Percentile
Interquartile Range
Standard Deviation Big shout out to