Hugging Face
Models
Datasets
Spaces
Buckets
new
Docs
Enterprise
Pricing
Website
Tasks
HuggingChat
Collections
Languages
Organizations
Community
Blog
Posts
Daily Papers
Learn
Discord
Forum
GitHub
Solutions
Team & Enterprise
Hugging Face PRO
Enterprise Support
Inference Providers
Inference Endpoints
Storage Buckets
Log In
Sign Up
Open to Collab
131.0
TFLOPS
1
21
42
Soumik Rakshit
geekyrakshit
Follow
WildGenie's profile picture
regisss's profile picture
rishiraj's profile picture
25 followers
·
40 following
http://geekyrakshit.dev
soumikRakshit96
soumik12345
soumikrakshit
AI & ML interests
Computer vision
Recent Activity
reacted
to
m-ric
's
post
with 👍
3 days ago
𝐇𝐮𝐠𝐠𝐢𝐧𝐠 𝐅𝐚𝐜𝐞 𝐫𝐞𝐥𝐞𝐚𝐬𝐞𝐬 𝐏𝐢𝐜𝐨𝐭𝐫𝐨𝐧, 𝐚 𝐦𝐢𝐜𝐫𝐨𝐬𝐜𝐨𝐩𝐢𝐜 𝐥𝐢𝐛 𝐭𝐡𝐚𝐭 𝐬𝐨𝐥𝐯𝐞𝐬 𝐋𝐋𝐌 𝐭𝐫𝐚𝐢𝐧𝐢𝐧𝐠 𝟒𝐃 𝐩𝐚𝐫𝐚𝐥𝐥𝐞𝐥𝐢𝐳𝐚𝐭𝐢𝐨𝐧 🥳 🕰️ Llama-3.1-405B took 39 million GPU-hours to train, i.e. about 4.5 thousand years. 👴🏻 If they had needed all this time, we would have GPU stories from the time of Pharaoh 𓂀: "Alas, Lord of Two Lands, the shipment of counting-stones arriving from Cathay was lost to pirates, this shall delay the building of your computing temple by many moons " 🛠️ But instead, they just parallelized the training on 24k H100s, which made it take just a few months. This required parallelizing across 4 dimensions: data, tensor, context, pipeline. And it is infamously hard to do, making for bloated code repos that hold together only by magic. 🤏 𝗕𝘂𝘁 𝗻𝗼𝘄 𝘄𝗲 𝗱𝗼𝗻'𝘁 𝗻𝗲𝗲𝗱 𝗵𝘂𝗴𝗲 𝗿𝗲𝗽𝗼𝘀 𝗮𝗻𝘆𝗺𝗼𝗿𝗲! Instead of building mega-training codes, Hugging Face colleagues cooked in the other direction, towards tiny 4D parallelism libs. A team has built Nanotron, already widely used in industry. And now a team releases Picotron, a radical approach to code 4D Parallelism in just a few hundred lines of code, a real engineering prowess, making it much easier to understand what's actually happening! ⚡ 𝗜𝘁'𝘀 𝘁𝗶𝗻𝘆, 𝘆𝗲𝘁 𝗽𝗼𝘄𝗲𝗿𝗳𝘂𝗹: Counting in MFU (Model FLOPs Utilization, how much the model actually uses all the compute potential), this lib reaches ~50% on SmolLM-1.7B model with 8 H100 GPUs, which is really close to what huge libs would reach. (Caution: the team is leading further benchmarks to verify this) Go take a look 👉 https://github.com/huggingface/picotron/tree/main/picotron
liked
a dataset
4 days ago
ILSVRC/imagenet-1k
upvoted
an
article
18 days ago
Profiling in PyTorch (Part 1): A Beginner's Guide to torch.profiler
View all activity
Organizations
spaces
4
Sort: Recently updated
Sleeping
Agents
Aryabhatta Inference
🌍
Sleeping
Line Art Data Annotation
🚀
An app for annotation of line art data from books
Runtime error
2
MedRAG Multi-Modal
🩺
Runtime error
1
Enhance Me
🌖
models
5
Sort: Recently updated
geekyrakshit/binary-classifier
67M
•
Updated
Nov 29, 2024
•
3
geekyrakshit/grays-anatomy-index-medcpt
Updated
Nov 3, 2024
•
2
geekyrakshit/grays-anatomy-index-contriever
Updated
Nov 3, 2024
•
5
geekyrakshit/grays-anatomy-index
Updated
Nov 3, 2024
•
4
geekyrakshit/DeepLabV3-Plus
Updated
Jul 3, 2023
•
105
datasets
22
Sort: Recently updated
geekyrakshit/art-images
Viewer
•
Updated
21 days ago
•
12.6k
•
42
geekyrakshit/hotpotqa_sft_traces
Viewer
•
Updated
Dec 15, 2025
•
5
•
19
geekyrakshit/issues
Viewer
•
Updated
Oct 17, 2025
•
1.89k
•
12
geekyrakshit/rust-issues
Viewer
•
Updated
Oct 10, 2025
•
1.89k
•
40
•
2
geekyrakshit/hyperswitch
Viewer
•
Updated
Oct 7, 2025
•
550
•
6
geekyrakshit/drawing-made-easy
Viewer
•
Updated
Aug 12, 2025
•
95
•
8
geekyrakshit/prompt-injection-dataset
Viewer
•
Updated
Nov 29, 2024
•
534k
•
304
•
8
geekyrakshit/test-chunk-dataset
Viewer
•
Updated
Nov 23, 2024
•
832
•
12
geekyrakshit/test-dataset
Viewer
•
Updated
Nov 23, 2024
•
22
•
9
geekyrakshit/indian-legal-acts
Viewer
•
Updated
Nov 19, 2024
•
2.38k
•
41
•
2
View 22 datasets