It's really hard to evaluate LLM applications The most direct way is to do so is to gather feedback from the end user Here's an in depth walkthrough of using LangSmith to do that. Feedback is associated with traces, so you can easily debug bad results https://
github.com/langchain-ai/l
angsmith-cookbook/tree/main/feedback-examples/streamlit
…
CODE
-
Evaluating LLM Applications with LangSmith Feedback
By
–
-
LangSmith Real-World Examples: From LLM Prototype to Production
By
–
More generally we're going to putting together a lot of real world, end-to-end examples of using LangSmith to bring LLM applications from prototype to production Any particular examples you'd like to see?
-
Replit Increases Storage Limits for Builder Accounts
By
–
Starting today, if you have Repls at or close to 1 GiB, you can now surpass those limits based on your account level. Replit builders will get the following account-wide storage limits: – Free: 10 GiB
– Hacker: 20 GiB
– Pro: 50 GiB -
Llama2-7B Now Runs on Replit with Boosted Machines
By
–
It's now easier than ever to start and stay on Replit no matter how big your code or filesystem gets – even if it’s a whole LLM.
— Replit ⠕ (@Replit) 8 août 2023
We put it to the test. Check out Llama2-7B running on Replit.
Try it yourself and add a boosted machine to the Repl for the best performance. pic.twitter.com/O6EZ1Au3oMIt's now easier than ever to start and stay on Replit no matter how big your code or filesystem gets – even if it’s a whole LLM. We put it to the test. Check out Llama2-7B running on Replit. Try it yourself and add a boosted machine to the Repl for the best performance.
-
Replit Expands Cloud Storage Capacity to 1 TiB
By
–
Expandable Storage on Replit is here. Starting today, Replit account storage can expand up to 1 TiB. Expand your project ideas and check out these big Repls you can now run on Replit
-
Text-to-SQL Deep Dive: Steps and Implementation Guide
By
–
A great deep dive by @manuelsoria_ and @RLanceMartin on text-to-SQL and all the steps involved!
-
Neural Layer Influence Patterns: Fine-grained Wording vs Semantics
By
–
On the other hand, individual influence queries show distinct influence patterns. The bottom and top layers seem to focus on fine-grained wording while middle layers reflect higher-level semantic information. (Here, rows correspond to layers and columns correspond to sequences.)
-
Analyzing Pretraining Finetuning Interactions and Neural Circuits
By
–
This work is just the beginning. We hope to analyze the interactions between pretraining and finetuning, and combine influence functions with mechanistic interpretability to reverse engineer the associated circuits. You can read more on our blog:
-
Influence Functions: Measuring Training Data Impact on Models
By
–
Influence functions are a classic technique from statistics. They are formulated as a counterfactual: if a copy of a given training sequence were added to the dataset, how would that change the trained parameters (and, by extension, the model’s outputs)?
-
Stability AI Launches StableCode Generative AI LLM for Developers
By
–
Exciting news! Stability AI has launched StableCode, the revolutionary generative AI LLM for coding! Developers, get ready to level up your coding game! #AI #Coding #StableCode #StabilityAI https://
bit.ly/44Z8iO0