INDEX

Explanations

references to terms prefixed with "TR"

oai_token-act-pair · gpt-3.5-turbo

references to the abbreviation "TR"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 actionGroup

-0.77

manship

-0.73

esville

-0.72

 arts

-0.70

holder

-0.69

eers

-0.68

hold

-0.66

furt

-0.65

lace

-0.64

comes

-0.62

POSITIVE LOGITS

ACK

0.96

umble

0.92

UTH

0.90

ractor

0.88

IP

0.84

acement

0.82

ACT

0.81

acing

0.81

idy

0.81

andom

0.80

Activations Density 0.005%

references to terms prefixed with "TR"

references to the abbreviation "TR"

No Comments

No Known Activations