INDEX

Explanations

names of individuals

oai_token-act-pair · gpt-3.5-turbo

names of individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

adobe

-0.66

milo

-0.66

-0.61

Cth

-0.60

isition

-0.59

Environment

-0.58

Redditor

-0.56

effic

-0.56

 THEM

-0.54

resy

-0.54

POSITIVE LOGITS

 alike

1.39

 respectively

1.19

 together

0.97

 jointly

0.87

together

0.79

selves

0.77

 selves

0.75

 mutually

0.75

 respective

0.75

are

0.75

Activations Density 0.211%

names of individuals

names of individuals

No Comments

No Known Activations