OPEN SOURCE

–

26 April 2023 23h55

The important piece of info to remember here is that we trained this model in under 10 days. The training was even called internally “YOLO RUN”. We’re excited about these benchmarks but can do even better. We've already started work to train a 7B model.

Replit Fine-Tuned Model Outperforms Codex with Better Efficiency

By

–

26 April 2023 23h55

Both models also benchmark impressively well against commercial models. replit-finetuned-v1-3b is by far the smallest model on the table and it outperformed Codex and LLaMA. PaLM-Coder is 200x larger and we’re closing in on their performance with much better latency.

Code Model Demonstrates Surprising Non-Coding Reasoning Capabilities

By

–

26 April 2023 23h55

We also noticed a surprising capability in non-coding reasoning, despite the model being trained entirely on code. We benchmarked against models trained for reasoning tasks and replit-code-v1-3b performed incredibly well.

Replit Open-Sources Complete Code Model 2.7B

By

–

26 April 2023 23h55

At #ReplitDevDay, we announced we’ve trained and are open-sourcing our first Complete Code model. Introducing replit-code-v1-3b: – 2.7B params
– 20 languages
– 525B tokens
– 40% better than comparable models
– Trained in 10 days Take a look at the benchmarks yourself

Replit Code Models Outperform Larger Open Source Competitors

By

–

26 April 2023 23h55

Replit-code-v1-3b & replit-finetuned-v1-3b were trained entirely on code and were meant for single-line code completion. We didn’t expect either to perform so well on HumanEval, but they did. replit-finetuned-v1-3b outperformed all OSS code models, even those 5x its size.

→ View original post on X — @databricks

ydata-profiling Now Supports Spark DataFrames at Scale

By

@databricks

–

26 April 2023 22h02

Now supporting Spark DataFrames, ydata-profiling (previously pandas-profiling) helps tackle data profiling needs at scale! Learn how to incorporate it into your Databricks Notebooks & data flows to leverage data profiling w/ just a few lines of code

→ View original post on X — @abhi1thakur

AutoTrain Advanced Now Supports LLM Finetuning

By

@abhi1thakur

–

26 April 2023 21h45

AutoTrain Advanced supports LLM finetuning now! More details come tomorrow! Stay tuned or try without documentation/walkthrough if you dare 😉

→ View original post on X — @aiatmeta

SSL Cookbook: Collaborative Research for Democratizing Self-Supervised Learning

By

@aiatmeta

–

26 April 2023 18h20

The SSL cookbook features insights from more than a dozen authors from @NYUTandon
, @umiacs
, @UCDavis
, @UMontreal + Meta AI researchers, such as @ylecun We’re excited to share this with the community as part of our effort to lower barriers + democratize access to SSL research.