INDEX

Explanations

the word 'ug' at varying activation levels

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of the term "Guggenheim."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

peak

-0.70

 Leilan

-0.69

 calming

-0.65

 compr

-0.62

cape

-0.61

 infancy

-0.61

 wards

-0.61

 blocker

-0.60

 targ

-0.60

 Hemp

-0.60

POSITIVE LOGITS

glers

1.42

uese

1.20

gery

1.15

uay

1.15

ged

1.08

ging

1.06

nant

1.00

gers

0.99

ger

0.96

ats

0.93

Activations Density 0.012%

the word 'ug' at varying activation levels

repeated mentions of the term "Guggenheim."

No Comments

No Known Activations