INDEX

Explanations

capitalized words or acronyms preceded by a 'W'

oai_token-act-pair · gpt-3.5-turbo

occurrences of the letter 'W'

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

 unpre

-0.80

 gratification

-0.74

arial

-0.72

 apprehension

-0.69

uate

-0.68

 afore

-0.67

 unarmed

-0.67

 succeeding

-0.66

 bottleneck

-0.63

Khe

-0.63

POSITIVE LOGITS

atts

1.31

OW

1.20

restling

1.19

atson

1.16

INGS

1.12

edge

1.12

reck

1.11

ITCH

1.10

ITNESS

1.09

ITH

1.08

Activations Density 0.037%

capitalized words or acronyms preceded by a 'W'

occurrences of the letter 'W'

No Comments

No Known Activations