INDEX

Explanations

the word "des" followed by a single character

oai_token-act-pair · gpt-3.5-turbo

terms related to the concept of 'destruction' or 'damaging' actions

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

OWS

-0.77

glers

-0.74

DAY

-0.74

Reviewer

-0.73

ancial

-0.72

hetti

-0.72

regor

-0.71

razil

-0.70

intendent

-0.67

ONY

-0.66

POSITIVE LOGITS

ync

0.96

ugar

0.91

ktop

0.90

erve

0.89

plet

0.85

perate

0.85

semb

0.84

irable

0.82

viron

0.81

erving

0.81

Activations Density 0.005%

the word "des" followed by a single character

terms related to the concept of 'destruction' or 'damaging' actions

No Comments

No Known Activations