INDEX

Explanations

words related to harmful or illegal activities

oai_token-act-pair · gpt-3.5-turbo

terms related to harmful or deceitful actions

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ĸļ

-1.14

arist

-0.87

ills

-0.79

ļéĨĴ

-0.79

hetti

-0.77

gdala

-0.77

ADRA

-0.72

oran

-0.72

ucket

-0.70

annis

-0.69

POSITIVE LOGITS

ly

0.94

 payload

0.90

 malicious

0.88

 intent

0.88

 mischief

0.79

 implant

0.78

vertising

0.78

icious

0.77

 behaviour

0.75

 behavi

0.71

Activations Density 0.006%

words related to harmful or illegal activities

terms related to harmful or deceitful actions

No Comments

No Known Activations