Yosef Worku Alemneh
rasyosef
AI & ML interests
Pretraining, Supervised Fine Tuning, Direct Preference Optimization, Retrieval Augmented Generation (RAG), Function Calling
Recent Activity
updated
a model
22 days ago
rasyosef/roberta-amharic-reranker-medium
updated
a collection
23 days ago
Llama 3.2 Amharic
Organizations
rasyosef's activity
Using hard negatives VS query, pos pair to train embedding models
4
#2 opened about 1 month ago
by
rasyosef
Adding Evaluation Results
#1 opened 7 months ago
by
leaderboard-pr-bot

Adding Evaluation Results
#3 opened 7 months ago
by
leaderboard-pr-bot

Phi-2-Instruct-APO: aligned with Anchored Preference Optimization
16
#3 opened 7 months ago
by
rasyosef
[Query-ISSUE] tokenizer.vocab_size is 128000, however len(tokenizer) is 128256, which prevents me from using those other tokens.
1
#34 opened 5 months ago
by
HV-Khurdula

What are the start and stop tokens of this model?
1
#40 opened 5 months ago
by
aryaash
Is the BOS token id of 128000 hardcoded into the llama 3.2 tokenizer?
2
#17 opened 6 months ago
by
rasyosef
Mistral-NeMo-Minitron-8B-Chat
5
#5 opened 8 months ago
by
rasyosef
APO Trainer in TRL?
1
#2 opened 7 months ago
by
rasyosef
ChatML template does not work properly
10
#2 opened 8 months ago
by
WasamiKirua

Collaboration
1
#1 opened 8 months ago
by
deleted
Error when trying to run
1
#1 opened 8 months ago
by
ctranslate2-4you
What changed for people using this model in english?
3
#3 opened 8 months ago
by
migueltalka
Phi 2 Instruct: an instruction following Phi 2 SLM that has undergone SFT and DPO
#132 opened 8 months ago
by
rasyosef
Phi 1.5 Instruct: an instruction following Phi 1.5 model that has undergone SFT and DPO
#89 opened 8 months ago
by
rasyosef
Update README.md
1
#2 opened 9 months ago
by
seyyaw

Duplicate?
1
#2 opened 11 months ago
by
israel
