AI & ML interests

None defined yet.

Recent Activity

Cyrile  updated a collection 27 days ago
ODQA
Cyrile  updated a dataset about 1 month ago
cmarkea/tab-doc-image
View all activity

cmarkea 's collections 6

DistilCamemBERT
DistilCamemBERT is a family of models derived from distilling the French model CamemBERT.
ODQA
Models based on Bloomz-Chat, specialized for subtasks of Open-Domain Question Answering, such as QA, RAG, etc., language-agnostic for both En and Fr.
Zero-shot classifier
Zero-shot classifiers (without training) are very useful for labeling or pre-labeling when there is little or no tagged data available.
Document Processing
Any model or dataset dealing with documentary-type objects (layout detection, VQA, OCR, etc.)
Quantized 4-bit models
Large model quantized with post-quantization performance very close to the original models, allowing it to run on reasonable infrastructure.
DistilCamemBERT
DistilCamemBERT is a family of models derived from distilling the French model CamemBERT.
ODQA
Models based on Bloomz-Chat, specialized for subtasks of Open-Domain Question Answering, such as QA, RAG, etc., language-agnostic for both En and Fr.
Document Processing
Any model or dataset dealing with documentary-type objects (layout detection, VQA, OCR, etc.)
Zero-shot classifier
Zero-shot classifiers (without training) are very useful for labeling or pre-labeling when there is little or no tagged data available.
Quantized 4-bit models
Large model quantized with post-quantization performance very close to the original models, allowing it to run on reasonable infrastructure.