unilm
involution
| unilm | involution | |
|---|---|---|
| 40 | 6 | |
| 18,407 | 1,306 | |
| 1.7% | - | |
| 9.0 | 0.0 | |
| 2 days ago | almost 3 years ago | |
| Python | Python | |
| MIT License | MIT License |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
unilm
- The Era of 1-Bit LLMs: Training_Tips, Code And_FAQ [pdf]
- The Era of 1-Bit LLMs: Training Tips, Code and FAQ
-
The Era of 1-bit LLMs: ternary parameters for cost-effective computing
+1 On this, the real proof would have been testing both models side-by-side.
It seems that it may be published on GitHub [1] according to HuggingFace [2].
[1] https://github.com/microsoft/unilm/tree/master/bitnet
[2] https://huggingface.co/papers/2402.17764
- I'm an Old Fart and AI Makes Me Sad
-
On building a semantic search engine
e5-mistral is essentially a distillation from gpt-4 to a smaller model. You can see here https://github.com/microsoft/unilm/blob/16da2f193b9c1dab0a69...
they actually have custom prompts for each dataset being tested.
Question would be, if you haven't seen the task before, what is a good prompt to prepend for your task?
IMO e5-mistral is overfit to MTEB
-
Leveraging GPT-4 for PDF Data Extraction: A Comprehensive Guide
Layout LM v1, v2 and v3 models [ Github ] DocBERT [ Github ]
-
Microsoft Publishes LongNet: Scaling Transformers to 1,000,000,000 Tokens
The repository is available here.
-
Recommended open LLMs with image input modality?
It is missing kosmos-2. I remember its image captioning was(demo currently down) really good and it's almost as fast as llava and lavin.
-
LongNet: Scaling Transformers to 1,000,000,000 Tokens
Should be this: https://github.com/microsoft/unilm/
-
[R] LongNet: Scaling Transformers to 1,000,000,000 Tokens
This is from Microsoft Research (Asia). https://aka.ms/GeneralAI
involution
-
[R] Involution: Inverting the Inherence of Convolution for Visual Recognition
PDF Link | Landing Page | Read as web page on arXiv Vanity
-
[P] PyTorch Involution layer wrapper
Have you benchmarked it against the official implementations? Would be interesting to see what the difference is versus their CUDA version.
-
[D] Paper Explained - Involution: Inverting the Inherence of Convolution for Visual Recognition (Full Video Analysis)
Code: https://github.com/d-li14/involution
What are some alternatives?
transformers - 🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
efficientdet-pytorch - A PyTorch impl of EfficientDet faithful to the original Google impl w/ ported weights
ERNIE - Official implementations for various pre-training models of ERNIE-family, covering topics of Language Understanding & Generation, Multimodal Understanding & Generation, and beyond.
mmdetection - OpenMMLab Detection Toolbox and Benchmark
gensim - Topic Modelling for Humans
Im2Vec - [CVPR 2021 Oral] Im2Vec Synthesizing Vector Graphics without Vector Supervision
maelstrom - A workbench for writing toy implementations of distributed systems.
yolact_edge - The first competitive instance segmentation approach that runs on small edge devices at real-time speeds.
rasa - 💬 Open source machine learning framework to automate text- and voice-based conversations: NLU, dialogue management, connect to Slack, Facebook, and more - Create chatbots and voice assistants
manydepth - [CVPR 2021] Self-supervised depth estimation from short sequences
memprompt - A method to fix GPT-3 after deployment with user feedback, without re-training.
MPRNet - [CVPR 2021] Multi-Stage Progressive Image Restoration. SOTA results for Image deblurring, deraining, and denoising.