INDEX

Explanations

numbers written in digits with the word 'digit'

oai_token-act-pair · gpt-3.5-turbo

occurrences of numerical figures, particularly those indicating counts or economic data

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

hire

-0.93

heres

-0.69

side

-0.68

haven

-0.68

deals

-0.64

nda

-0.63

 councils

-0.63

nd

-0.62

dep

-0.62

sterdam

-0.61

POSITIVE LOGITS

 digit

1.32

digit

1.05

 digits

1.01

igr

0.93

ized

0.93

omial

0.93

itized

0.89

ocene

0.88

puter

0.87

alogue

0.85

Activations Density 0.010%

numbers written in digits with the word 'digit'

occurrences of numerical figures, particularly those indicating counts or economic data

No Comments

No Known Activations