Dimitris Roussis
droussis
AI & ML interests
All things data for LLMs, NMT, evaluation, safety, alignment, and more
Recent Activity
liked
a dataset
11 days ago
QubitPi/wiktionary-data
upvoted
an
article
16 days ago
Releasing the largest multilingual open pretraining dataset
New activity
21 days ago
lightblue/rag_multilingual_training_negatives:Is this the same as Kurage?
Organizations
droussis's activity
Is this the same as Kurage?
2
#2 opened 25 days ago
by
droussis
Context extension?
1
#4 opened 2 months ago
by
droussis
Dataset fails loading
#2 opened 3 months ago
by
droussis
Dataset fails loading
#1 opened 4 months ago
by
droussis
About context size and difference in quality
3
#1 opened 7 months ago
by
droussis
Future plans (Llama 3?)
1
#3 opened 7 months ago
by
velocity
LLama-Factory inference issue
14
#2 opened 8 months ago
by
ianss
Regarding quality assessment
2
#1 opened 11 months ago
by
droussis
Community request: more languages
8
#1 opened 11 months ago
by
emre
Which part of HC3?
#1 opened about 1 year ago
by
droussis
Difference between 40K and 395K dataset
#2 opened about 1 year ago
by
droussis
The model output is totally corrupted
4
#5 opened over 1 year ago
by
fernandofernandes
Fix weights by putting the right value in `lm_head.weight`
3
#3 opened over 1 year ago
by
sgugger
The model output is totally corrupted
4
#5 opened over 1 year ago
by
fernandofernandes