bartowski/agentica-org_DeepScaleR-1.5B-Preview-GGUF
Text Generation
•
Updated
•
54.5k
•
38
Serge Brun
surfiend
·
AI & ML interests
None yet
Recent Activity
liked
a model
23 days ago
bartowski/agentica-org_DeepScaleR-1.5B-Preview-GGUF
liked
a Space
about 1 month ago
ggml-org/gguf-my-repo
liked
a model
about 1 month ago
deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B
Organizations
None yet
surfiend's activity
liked
a
model
23 days ago
liked
a
Space
about 1 month ago
liked
a
model
about 1 month ago
updated
a
collection
5 months ago
upvoted
2
articles
5 months ago
Article
Fine-tuning LLMs to 1.58bit: extreme quantization made easy
•
225
Article
Document Similarity Search with ColPali
By
•
•
49
upvoted
a
paper
6 months ago
liked
a
model
7 months ago
liked
a
Space
8 months ago
liked
a
model
10 months ago
upvoted
a
paper
11 months ago
liked
2
models
12 months ago
liked
a
model
about 1 year ago
upvoted
2
papers
about 1 year ago
liked
a
Space
about 1 year ago
reacted to
osanseviero's
post with 👍
about 1 year ago
Post
Mixture of experts: beware 🛡️⚔️
New paper by DeepMind: Buffer Overflow in MoE Buffer Overflow in Mixture of Experts (2402.05526)
The paper shows an adversarial attack strategy in which a user sends malicious queries that can affect the output of other user queries from the same batch.
So if in the same batch we have
- User A benign query
- User B malicious query
The response for A might be altered!😱
How is this possible?
One approach is to fill the token buffers with adversarial data, hence forcing the gating to use the non-ideal experts or to entirely drop the bening tokens (in the case of finite limit size).
This assumes that the adversary can use the model as a black-box but can observe the logit outputs + ensure that the data is always grouped in the same batch.
How to mitigate this?
- Randomize batch order (and even run twice if some queries are very sensitive)
- Use a large capacity slack
- Sample from gate weights instead of top-k (not great IMO, as that require more memory for inference)
Very cool paper!!
New paper by DeepMind: Buffer Overflow in MoE Buffer Overflow in Mixture of Experts (2402.05526)
The paper shows an adversarial attack strategy in which a user sends malicious queries that can affect the output of other user queries from the same batch.
So if in the same batch we have
- User A benign query
- User B malicious query
The response for A might be altered!😱
How is this possible?
One approach is to fill the token buffers with adversarial data, hence forcing the gating to use the non-ideal experts or to entirely drop the bening tokens (in the case of finite limit size).
This assumes that the adversary can use the model as a black-box but can observe the logit outputs + ensure that the data is always grouped in the same batch.
How to mitigate this?
- Randomize batch order (and even run twice if some queries are very sensitive)
- Use a large capacity slack
- Sample from gate weights instead of top-k (not great IMO, as that require more memory for inference)
Very cool paper!!
- 621 replies
liked
a
model
about 1 year ago
