Fragile Knowledge, Robust Instruction-Following: The Width Pruning Dichotomy in Llama-3.2
Paper
•
2512.22671
•
Published
•
1
Pere Martra
oopere
AI & ML interests
ML Research Engineer | LLM Optimization & Efficiency | Authoring “Rearchitecting LLMs” @Manning | 1.9K GitHub ⭐
Recent Activity
updated
a model
2 days ago
oopere/gemma-3-270m-14L-distilled
published
a model
2 days ago
oopere/gemma-3-270m-14L-distilled
liked
a dataset
3 days ago
gmongaras/SlimPajama-627B_Reupload
Organizations
upvoted
a
paper
about 1 month ago
upvoted
an
article
3 months ago
Article
Making LLMs Smaller Without Breaking Them: A GLU-Aware Pruning Approach
•
18
upvoted
2
articles
7 months ago
Article
Good answers are not necessarily factual answers: an analysis of hallucination in leading LLMs
•
42
Article
LLMs recognise bias but also reproduce harmful stereotypes: an analysis of bias in leading LLMs
•
16
upvoted
an
article
8 months ago
Article
Welcome Gemma 3: Google's all new multimodal, multilingual, long context open LLM
- +2
•
485
upvoted
an
article
11 months ago
Article
Training a Gemma 2 2B-IT for Reasoning with GRPO
•
6
upvoted
an
article
about 1 year ago
Article
Open-R1: a fully open reproduction of DeepSeek-R1
- +1
•
887
upvoted
a
paper
over 1 year ago
upvoted
2
articles
over 1 year ago
Article
Preference Tuning LLMs with Direct Preference Optimization Methods
- +3
•
78
Article
How NuminaMath Won the 1st AIMO Progress Prize
- +6
•
126
upvoted
a
collection
over 1 year ago
upvoted
an
article
over 1 year ago
Article
🪆 Introduction to Matryoshka Embedding Models
- +1
•
192
upvoted
an
article
almost 2 years ago
Article
Making LLMs even more accessible with bitsandbytes, 4-bit quantization and QLoRA
- +3
•
173