INDEX

Explanations

names of individuals, likely with some level of significance or fame

oai_token-act-pair · gpt-3.5-turbo

names of notable individuals or works

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

 undermin

-0.60

 corrid

-0.59

 predec

-0.59

 equivalents

-0.56

acebook

-0.55

illary

-0.53

akespe

-0.52

aturdays

-0.52

Versions

-0.52

 cryst

-0.51

POSITIVE LOGITS

(@

0.63

 Profile

0.58

zinski

0.58

gaard

0.54

 Originally

0.53

han

0.52

agh

0.52

fried

0.52

zer

0.52

 Posted

0.52

Activations Density 0.539%

names of individuals, likely with some level of significance or fame

names of notable individuals or works

No Comments

No Known Activations