4/ ChatDoctor – a medical chat model fine-tuned on LLaMA using medical domain knowledge. Collects data on around 700 diseases and generated 5K doctor-patient conversations to finetune the LLM.
OPEN SOURCE
-
LLaMA-Adapter: Lightweight Method Fine-Tune Instruction-Following Models
By
–
5/ LLaMA-Adapter – a lightweight adaption method to efficiently fine-tune LLaMA into an instruction-following model; generates responses comparable to Alpaca with fully fine-tuned 7B parameter; it’s also extended for multi-modal input support.
-
AI Training Data Crisis: Documentation and GPL Solutions
By
–
Was discussing this with @monkchips
. "Docs" likely to increasingly look like "ChatGPT", but @peternixey raises a critical point: where will the training data come from if devs are quietly asking machines for guidance? Maybe we need a new GPL to avoid one-way knowledge traps? -
Store-bought versus homemade: choosing practical or custom solutions
By
–
I go back and forth. It’s like store bought vs homemade bbq sauce. Store bought is quick and great (more practical for me), home made is slow, tedious, and not always great – but when it’s great, it’s the best.
-
RWKV: RNN-based LLM combining RNNs and Transformers capabilities
By
–
RWVK: an #RNN-based LLM Feels like the MLP-Mixer moment for LLMs, combining the best of RNNs (longer sequences) and #Transformers (parallelization). The Raven chatbot is impressive and on par with recent models like GPT4All. https://
github.com/BlinkDL/RWKV-LM -
French Language Dataset Size and LLM Training Data Deduplication
By
–
ça fait que 32go a télécharger le fr avec les images. C'était le cas pour THE PILE qui a servi de base à d'autre LLM comme GPT-J-6 ou GPT-neo, nlg-megatron, mais me semble pas. A moins qu'ils aient considére que ça faisait doublon et que la trad serait suffisante. Tu peux
-
Integrating Data Processing Workflow with Hugging Face Ecosystem
By
–
Very cool! Congratulations Is the codebase already online somewhere? Would be cool to see if we can incorporate your data processing workflow using the @huggingface ecosystem!
-
Analysis of Twitter’s open-sourced recommendation algorithm
By
–
[THREAD] J'analyse l'#algorithme de recommendation de #Twitter qui vient d'être open sourcé. Thread directement sur le compte @deeplayerAI
-
Cornell CS5785 Applied Machine Learning Course with Jupyter Notebooks
By
–
Applied Machine Learning (Cornell CS5785) "Starting from the very basics, covering all of the most important ML algorithms and how to apply them in practice. Executable Jupyter notebooks (and as slides)". Lectures: https://
youtube.com/playlist?list=
PL2UML_KCiC0UlY7iCQDSiGDMovaupqc83
…
Notebooks: https://
github.com/kuleshov/corne
ll-cs5785-2020-applied-ml
… -
Hub Releases New AI Datasets for Community
By
–
Oh actually nvm they did release some datasets on the Hub