Dimitris Roussis's picture

Dimitris Roussis

droussis

·

https://www.ilsp.gr/en/members/roussis-dimitris/

droussis

AI & ML interests

All things data for LLMs, NMT, evaluation, safety, alignment, and more

Recent Activity

updated a dataset about 2 hours ago

ilsp/flores200_el-x

published a dataset about 2 hours ago

ilsp/flores200_el-x

liked a dataset 4 days ago

mxersion/MX-CHAT

View all activity

Organizations

droussis's activity

New activity in lightblue/rag_multilingual_training_negatives 3 months ago

Is this the same as Kurage?

#2 opened 3 months ago by

New activity in utter-project/EuroLLM-1.7B-Instruct 4 months ago

Context extension?

#4 opened 4 months ago by

New activity in PleIAs/Greek-PD 5 months ago

Dataset fails loading

#2 opened 5 months ago by

New activity in PleIAs/WTO-Text 6 months ago

Dataset fails loading

#1 opened 6 months ago by

New activity in bitextor/bicleaner-ai-full-large-en-xx 9 months ago

About context size and difference in quality

#1 opened 9 months ago by

New activity in ilsp/Meltemi-7B-v1 9 months ago

Future plans (Llama 3?)

#3 opened 9 months ago by

New activity in ilsp/Meltemi-7B-Instruct-v1 10 months ago

LLama-Factory inference issue

#2 opened 10 months ago by

New activity in CausalLM/GPT-4-Self-Instruct-Greek about 1 year ago

Regarding quality assessment

#1 opened about 1 year ago by

New activity in CausalLM/GPT-4-Self-Instruct-German about 1 year ago

Community request: more languages

#1 opened about 1 year ago by

New activity in Intel/neural-chat-dataset-v2 about 1 year ago

Which part of HC3?

#1 opened about 1 year ago by

New activity in meta-math/MetaMathQA-40K about 1 year ago

Difference between 40K and 395K dataset

#2 opened about 1 year ago by

New activity in Helsinki-NLP/opus-mt-tc-big-en-pt over 1 year ago

The model output is totally corrupted

#5 opened over 1 year ago by

fernandofernandes

New activity in Helsinki-NLP/opus-tatoeba-en-ja over 1 year ago

Fix weights by putting the right value in `lm_head.weight`

#3 opened over 1 year ago by

New activity in Helsinki-NLP/opus-mt-tc-big-en-pt over 1 year ago

The model output is totally corrupted

#5 opened over 1 year ago by

fernandofernandes