Róbert Belanec's picture

In a Training Loop 🔄

8 4

Róbert Belanec PRO

rbelanec

·

https://belanec.eu/

AI & ML interests

Parameter-Efficient Fine-Tuning, Multi-Task Transfer-Learning, Model Merging, Efficient Training

Recent Activity

updated a model about 12 hours ago

rbelanec/train_record_456_1766622579

published a model 2 days ago

rbelanec/train_record_456_1766622579

updated a model 9 days ago

rbelanec/train_record_456_1765807514

View all activity

Organizations

upvoted a collection about 1 month ago

PEFT-Factory

Collection of datasets adapted and preprocessed to be compatible with PEFT-Factory • 14 items • Updated Nov 14 • 1

upvoted 3 papers 3 months ago

Fighting Randomness with Randomness: Mitigating Optimisation Instability of Fine-Tuning using Delayed Ensemble and Noisy Interpolation

Paper • 2406.12471 • Published Jun 18, 2024 • 1

Controlling the Output of a Generative Model by Latent Feature Vector Shifting

Paper • 2311.08850 • Published Nov 15, 2023 • 1

Task Prompt Vectors: Effective Initialization through Multi-Task Soft-Prompt Transfer

Paper • 2408.01119 • Published Aug 2, 2024 • 1

upvoted a collection 5 months ago

PEFT-Factory

Collections of datasets adapted for PEFT-Factory benchmark evaluation. • 14 items • Updated Nov 14 • 1

upvoted a collection 9 months ago

MMLU Paraphrases

Paraphrases of the max 500 tokens subset of the MMLU dataset. We train models on both paraphrases and not paraphrases to increase robustness. • 5 items • Updated Mar 18 • 1

upvoted an article 10 months ago

Article

Welcome to Inference Providers on the Hub 🔥

+5

Jan 28

•

491

upvoted a collection about 1 year ago

Task Prompt Vectors

Collection of all soft-prompts used in the paper https://arxiv.org/abs/2408.01119 • 16 items • Updated Mar 18 • 2