INDEX

Explanations

dates and specific details mentioned in emails or official communications

oai_token-act-pair · gpt-3.5-turbo

instances of the word "on" indicating dates and events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 endif

-0.82

hello

-0.66

 kissed

-0.66

hatt

-0.65

endif

-0.64

hole

-0.64

 kidding

-0.63

AMI

-0.63

African

-0.63

natureconservancy

-0.62

POSITIVE LOGITS

mbuds

0.79

erity

0.75

edom

0.71

raq

0.68

 unprecedented

0.68

itech

0.65

ioch

0.65

igible

0.65

spect

0.64

 December

0.63

Activations Density 0.396%

dates and specific details mentioned in emails or official communications

instances of the word "on" indicating dates and events

No Comments

No Known Activations