Moulay Amine Jaidi

208 posts

Moulay Amine Jaidi banner
Moulay Amine Jaidi

Moulay Amine Jaidi

@Semantichasm

Independent AI consultancy that specialise in optimising Machine Learning Workloads on Bare metal and cloud. Specialists in AMD #MI300X Multi-GPU optimisation.

Bristol Katılım Ekim 2024
44 Takip Edilen34 Takipçiler
Moulay Amine Jaidi
Moulay Amine Jaidi@Semantichasm·
This work is part of our broader effort to build a framework for testing AI systems. As LLMs become increasingly complex, it's crucial we use them safely and responsibly.
English
0
0
0
12
Moulay Amine Jaidi
Moulay Amine Jaidi@Semantichasm·
🚀 Jailbreaking LLMs Our recent research has explored how #adversarialai can be used to bypass refusal responses. Using @HotAisle Solo @AIatAMD #MI300X, we've auto-assessed #LLM using various approaches and successfully circumvented model guardrails of open-source models.
Moulay Amine Jaidi tweet mediaMoulay Amine Jaidi tweet media
English
3
0
1
724
Moulay Amine Jaidi
Moulay Amine Jaidi@Semantichasm·
The example below demonstrates how automated approaches can compel models to comply with requests they'd normally refuse when asked directly.
English
0
0
0
10
Karthik
Karthik@karthikponna19·
which one do you prefer for coding ?
Karthik tweet mediaKarthik tweet mediaKarthik tweet mediaKarthik tweet media
English
957
122
2.4K
235.7K
Tokyo
Tokyo@otokyo__·
Can you name ONE?
Tokyo tweet media
English
3.7K
86
977
312.5K
Elma
Elma@oelma__·
What is the first thing that comes to mind when you think of 🏴
Elma tweet media
English
3.9K
128
1.1K
152.6K
Nimrat
Nimrat@_nimmiikaur·
Men, delete one forever
Nimrat tweet media
English
4.4K
418
7.3K
15.6M
Tokyo
Tokyo@otokyo__·
One word?
Tokyo tweet media
English
1.7K
58
594
62.3K
Genius Tech
Genius Tech@Geniustechw·
Please tell me
Genius Tech tweet media
English
453
19
179
28.6K
Priyanka Lakhara
Priyanka Lakhara@codewithpri·
Which one do you use - Vscode or Cursor?
Priyanka Lakhara tweet mediaPriyanka Lakhara tweet media
English
737
76
1.7K
170.5K
Val Kantchev
Val Kantchev@valkantchev·
@SemiAnalysis_ Another thing that's missing is that AMD's software stack is a big joke!
English
1
0
0
625
SemiAnalysis
SemiAnalysis@SemiAnalysis_·
At AMD's Analyst Day this week, AMD's Head of AI show that on InferenceMAX, MI355X has 3-10x better perf per TCO than MI300/MI325 on GPTOSS 120B. One thing missing from their slide is the the classic obnoxiously giant SemiAnalysis logo
English
6
16
218
27K
Moulay Amine Jaidi
Moulay Amine Jaidi@Semantichasm·
@zzznah It's a good thing - papers are supposed to be peer reviewed and the code is hardly explored in details when they're submitted plus the author was reponsive
English
0
0
0
20
Alex Mordvintsev
Alex Mordvintsev@zzznah·
Ok, my complaint got too much attention so I guess I have to make a few comments. - In general I think the results look quite interesting and the code in github repo is more consistent than the example below - I'm not sure I buy the idea of a particular embedding distribution to be universally optimal in a way that makes all the difference, yet - the paper also suggests a nice backprop-friendly way to prevent embedding collapse, which I think is the key. I happened to miss "Epps-Pulley" test in and the idea ECFs in from my toolbox. I used to mostly rely on sorting based sliced-Wasserstein or Sinkhorn when I needed to match two distributions, so it's nice to have this education gap closed
Alex Mordvintsev@zzznah

me: read the paper... bad at math, just show me the code the code:

English
4
1
77
21.7K
Cody Blakeney
Cody Blakeney@code_star·
We are looking for a post-training lead at @datologyai we have gpus, you can make them go brrrr
Cody Blakeney tweet media
English
9
17
133
49.9K