contextualized-topic-models
transformers
Our great sponsors
contextualized-topic-models | transformers | |
---|---|---|
7 | 175 | |
1,163 | 125,021 | |
1.7% | 3.1% | |
5.0 | 10.0 | |
3 months ago | 5 days ago | |
Python | Python | |
MIT License | Apache License 2.0 |
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.
contextualized-topic-models
-
[Project]Topic modelling of tweets from the same user
In our experiments, CTM works well with tweets: https://github.com/MilaNLProc/contextualized-topic-models (I'm one of the authors)
-
Extract words from large data set of reviews by sentiment
Use CTM https://github.com/MilaNLProc/contextualized-topic-models with sentiment labels to built distribution of words over labels
-
Using Transformer for Topic Modeling - what are the options?
This library from MILA seems quite neat! I haven’t had the change to play with it though : https://github.com/MilaNLProc/contextualized-topic-models
-
Catogorize the Data- Topic Modelling algorithm
a bit of shameless self-promotion, but we developed a topic model (https://github.com/MilaNLProc/contextualized-topic-models) that actually supports that use case!
-
(NLP) Best practices for topic modeling and generating interesting topics?
If you use CTM, you can provide the topic model two inputs: the preprocessed texts (that will be used by the topic model to generate the topical words) and the unpreprocessed texts (to generate the contextualized representations that will be later concatenated to the document bag-of-word representation). We saw that this slightly improves the performance instead of providing BERT the already-preprocessed text. This feature is supported in the original implementation of CTM, not in OCTIS. See here: https://github.com/MilaNLProc/contextualized-topic-models#combined-topic-model
-
Latest trends in topic modelling?
Cross-lingual Contextualized Topic Models with Zero-shot Learning from a team at MilaNLP which uses bag of words representations in combination with multi lingual embeddings from SBERT and works like a VAE (encode the input, use the encoded representation to decode back to a bag of words as close to the input as possible). Using SBERT embeddings makes their model generalise for other languages which may be useful. One major shortfall of this model as I understand is that it can't deal with long documents very elegantly - only up to BERT'S word limit (the workaround is to truncate and use the first words)
transformers
-
Maxtext: A simple, performant and scalable Jax LLM
Is t5x an encoder/decoder architecture?
Some more general options.
The Flax ecosystem
https://github.com/google/flax?tab=readme-ov-file
or dm-haiku
https://github.com/google-deepmind/dm-haiku
were some of the best developed communities in the Jax AI field
Perhaps the “trax” repo? https://github.com/google/trax
Some HF examples https://github.com/huggingface/transformers/tree/main/exampl...
Sadly it seems much of the work is proprietary these days, but one example could be Grok-1, if you customize the details. https://github.com/xai-org/grok-1/blob/main/run.py
-
Lossless Acceleration of LLM via Adaptive N-Gram Parallel Decoding
The HuggingFace transformers library already has support for a similar method called prompt lookup decoding that uses the existing context to generate an ngram model: https://github.com/huggingface/transformers/issues/27722
I don't think it would be that hard to switch it out for a pretrained ngram model.
-
AI enthusiasm #6 - Finetune any LLM you wantđź’ˇ
Most of this tutorial is based on Hugging Face course about Transformers and on Niels Rogge's Transformers tutorials: make sure to check their work and give them a star on GitHub, if you please ❤️
-
Schedule-Free Learning – A New Way to Train
* Superconvergence + LR range finder + Fast AI's Ranger21 optimizer was the goto optimizer for CNNs, and worked fabulously well, but on transformers, the learning rate range finder sadi 1e-3 was the best, whilst 1e-5 was better. However, the 1 cycle learning rate stuck. https://github.com/huggingface/transformers/issues/16013
-
Gemma doesn't suck anymore – 8 bug fixes
Thanks! :) I'm pushing them into transformers, pytorch-gemma and collabing with the Gemma team to resolve all the issues :)
The RoPE fix should already be in transformers 4.38.2: https://github.com/huggingface/transformers/pull/29285
My main PR for transformers which fixes most of the issues (some still left): https://github.com/huggingface/transformers/pull/29402
- HuggingFace Transformers: Qwen2
- HuggingFace Transformers Release v4.36: Mixtral, Llava/BakLlava, SeamlessM4T v2
- HuggingFace: Support for the Mixtral Moe
-
Paris-Based Startup and OpenAI Competitor Mistral AI Valued at $2B
If you want to tinker with the architecture Hugging Face has a FOSS implementation in transformers: https://github.com/huggingface/transformers/blob/main/src/tr...
If you want to reproduce the training pipeline, you couldn't do that even if you wanted to because you don't have access to thousands of A100s.
-
Fail to reproduce the same evaluation metrics score during inference.
I am aware that using mixed precision reduces the stability of weight and there will be little consistency but don't expect it to be this much. I have attached the graph of evaluation metrics. If someone can give me some insight into this issue, that would be great.
What are some alternatives?
BERTopic - Leveraging BERT and c-TF-IDF to create easily interpretable topics.
fairseq - Facebook AI Research Sequence-to-Sequence Toolkit written in Python.
OCTIS - OCTIS: Comparing Topic Models is Simple! A python package to optimize and evaluate topic models (accepted at EACL2021 demo track)
sentence-transformers - Multilingual Sentence & Image Embeddings with BERT
PolyFuzz - Fuzzy string matching, grouping, and evaluation.
llama - Inference code for Llama models
tika-python - Tika-Python is a Python binding to the Apache Tika™ REST services allowing Tika to be called natively in the Python community.
transformer-pytorch - Transformer: PyTorch Implementation of "Attention Is All You Need"
Top2Vec - Top2Vec learns jointly embedded topic, document and word vectors.
text-generation-webui - A Gradio web UI for Large Language Models. Supports transformers, GPTQ, AWQ, EXL2, llama.cpp (GGUF), Llama models.
Sentimentanalysis - Language independent sentiment analysis
huggingface_hub - The official Python client for the Huggingface Hub.