INDEX

Explanations

mentions of unauthorized or illegal activities

oai_token-act-pair · gpt-3.5-turbo

references to unauthorized access or connections

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

=-=-=-=-=-=-=-=-

-1.08

=-=-=-=-

-0.89

utra

-0.79

achine

-0.78

mom

-0.77

oran

-0.76

 Dynamics

-0.76

hetti

-0.73

ills

-0.72

anches

-0.72

POSITIVE LOGITS

 unauthorized

0.87

 access

0.81

 disclosures

0.80

 reuse

0.74

 intruder

0.73

 disclosure

0.72

 permission

0.72

 interference

0.70

 downloading

0.69

 aggress

0.68

Activations Density 0.010%

mentions of unauthorized or illegal activities

references to unauthorized access or connections

No Comments

No Known Activations