Simo Ryu(@cloneofsimo) 's Twitter Profile Photo

YES!!!! TOOK 26 hours to make this happen: conditional D3PM implementation with pytorch. Let's accelerate discrete diffusion research!!! 👏I believe this is the only torch implementation of it out there!
Less than 400 LOC!
github.com/cloneofsimo/d3…
paper: arxiv.org/abs/2107.03006

YES!!!! TOOK 26 hours to make this happen: conditional D3PM implementation with pytorch. Let's accelerate discrete diffusion research!!! 👏I believe this is the only torch implementation of it out there!
Less than 400 LOC!
github.com/cloneofsimo/d3…
paper: arxiv.org/abs/2107.03006
account_circle
Normal Computing 🧠🌡️(@NormalComputing) 's Twitter Profile Photo

We're excited to announce posteriors!

posteriors is an open-source Python library designed to make it as easy as possible to apply uncertainty quantification to deep learning models with PyTorch.

github.com/normal-computi…

account_circle
Agippo(@Agippo_Vermith) 's Twitter Profile Photo

Does an Entity Relationship Diagram makes sense for a simple PyTorch chatbot (Flask) app that answers customer inquiries? Can one still use an ERD to model such an interaction? It’s only User, Inquiry, Intents (tags, tags, patterns, potential responses) & 1 final response🧐

account_circle
Patrick Loeber(@patloeber) 's Twitter Profile Photo

One of the YouTube videos I'm most proud. 730K people watched it🤯

Deep Learning With PyTorch - Full Course

However, it's 3 years old. Should I create a fresh one?

One of the YouTube videos I'm most proud. 730K people watched it🤯

Deep Learning With PyTorch - Full Course

However, it's 3 years old. Should I create a fresh one?
account_circle
Rohan Paul(@rohanpaul_ai) 's Twitter Profile Photo

Brilliant work from PyTorch team on releasing torchtune ✨

Running fine-tuning with single command.

Single-GPU recipes expose a number of memory optimizations that aren't available in the distributed versions.

torchtune is built with extensibility and usability, focussing on…

Brilliant work from @PyTorch  team on releasing torchtune ✨

Running fine-tuning with single command.

Single-GPU recipes expose a number of memory optimizations that aren't available in the distributed versions.

torchtune is built with extensibility and usability, focussing on…
account_circle
Om Alve(@alve_om) 's Twitter Profile Photo

Implemented LoRA in Pytorch and finetuned DistilBert on the IMDb reviews dataset by Stanford, the results are in the thread below

Implemented LoRA in Pytorch and finetuned DistilBert on the IMDb reviews dataset by Stanford, the results are in the thread below
account_circle
Rohan Paul(@rohanpaul_ai) 's Twitter Profile Photo

Nice PyTorch implementation of the landmark 'Infini-Transformer' paper in this repo. ✨

Instead of discarding the old key-value (KV) states like in standard attention, Infini-attention stores them in a compressive memory. When processing subsequent sequences, the values are…

Nice PyTorch implementation of the landmark 'Infini-Transformer' paper in this repo. ✨

Instead of discarding the old key-value (KV) states like in standard attention, Infini-attention stores them in a compressive memory. When processing subsequent sequences, the values are…
account_circle
Andrej Karpathy(@karpathy) 's Twitter Profile Photo

A few new CUDA hacker friends joined the effort and now llm.c is only 2X slower than PyTorch (fp32, forward pass) compared to 4 days ago, when it was at 4.2X slower 📈

The biggest improvements were:
- turn on TF32 (NVIDIA TensorFLoat-32) instead of FP32 for matmuls. This is a…

A few new CUDA hacker friends joined the effort and now llm.c is only 2X slower than PyTorch (fp32, forward pass) compared to 4 days ago, when it was at 4.2X slower 📈

The biggest improvements were:
- turn on TF32 (NVIDIA TensorFLoat-32) instead of FP32 for matmuls. This is a…
account_circle
Ben Victoria(@benvictoria17) 's Twitter Profile Photo

11. Training Your First Embedding Model
Training an embedding model can start with pre-existing libraries like TensorFlow or PyTorch. Beginners can use these tools to experiment with custom embeddings, adjusting parameters and architectures to suit their specific needs and data.

account_circle
Kartikay Khandelwal(@kakemeister) 's Twitter Profile Photo

Really excited to officially release torchtune: a PyTorch-native library for easily fine-tuning LLMs!

Code: github.com/pytorch/torcht…
Blog: pytorch.org/blog/torchtune…
Tutorials: pytorch.org/torchtune/stab…

[1/5]

account_circle
Tom Huang(@tuturetom) 's Twitter Profile Photo

RAG 基础数据质量处理组件 -- 自动发现数据问题并进行修复,提高数据质量和价值的 Cleanlab 开源,8.4K Star ⭐️

1. 支持任何数据,包括图片、文本、音频、表格等
2. 检测数据标注问题、改进、训练模型、价值提升💥
3. 支持任何框架或模型:OpenAI、HuggingFace、PyTorch

github.com/cleanlab/clean…

account_circle
Hubert Łępicki ❌🦶🐍(@hubertlepicki) 's Twitter Profile Photo

Oh I was able to set up ROCm without any problems with Python and PyTorch and it's currently running a model all within hour and no problems so far.

Looks like AMD started to get their shit together?

account_circle