INDEX

Explanations

instances of requiring login or being logged into an account

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "logged" and related terms connected to accessing accounts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

silver

-0.71

terday

-0.66

 Boris

-0.65

atown

-0.64

riages

-0.63

nces

-0.63

 Galile

-0.60

xual

-0.59

ansky

-0.59

ners

-0.59

POSITIVE LOGITS

gers

1.12

istically

1.08

 logs

0.99

istics

0.99

istical

0.91

ging

0.90

otype

0.86

itech

0.85

ophon

0.85

 logger

0.84

Activations Density 0.009%

instances of requiring login or being logged into an account

occurrences of the word "logged" and related terms connected to accessing accounts

No Comments

No Known Activations