INDEX

Explanations

phrases related to postal services or communication

oai_token-act-pair · gpt-3.5-turbo

mentions of specific organizations or acronyms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

stood

-0.71

practice

-0.68

pron

-0.67

flies

-0.64

definition

-0.64

ivities

-0.64

regular

-0.63

dimension

-0.63

 à¨

-0.62

ministic

-0.61

POSITIVE LOGITS

PO

1.20

etry

0.94

PO

0.94

arty

0.83

etary

0.83

INT

0.83

OTUS

0.82

ople

0.82

ODUCT

0.81

acher

0.78

Activations Density 0.007%

phrases related to postal services or communication

mentions of specific organizations or acronyms

No Comments

No Known Activations