INDEX

Explanations

the word "Gener" with varying suffixes and prefixes

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "generation" or generational topics

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

ĺħ

-0.82

mare

-0.72

laus

-0.69

 McKay

-0.66

 Sandwich

-0.63

ded

-0.62

wal

-0.62

 Sands

-0.62

foot

-0.60

uku

-0.60

POSITIVE LOGITS

ational

1.29

icity

1.25

ators

1.16

ative

1.15

ator

1.09

osity

1.08

andom

0.98

ations

0.96

andum

0.95

ating

0.94

Activations Density 0.017%

the word "Gener" with varying suffixes and prefixes

references to the concept of "generation" or generational topics

No Comments

No Known Activations