INDEX

Explanations

dates, particularly referencing periods in history

oai_token-act-pair · gpt-3.5-turbo

phrases indicating historical time periods

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

edIn

-0.79

dain

-0.70

henko

-0.68

illon

-0.66

 violates

-0.65

 captcha

-0.64

 actionGroup

-0.63

 sinks

-0.62

atches

-0.62

lifting

-0.62

POSITIVE LOGITS

est

0.98

liest

0.78

 Autumn

0.77

 twenties

0.77

0.75

-'

0.75

0.74

 sevent

0.72

era

0.72

Activations Density 0.057%

dates, particularly referencing periods in history

phrases indicating historical time periods

No Comments

No Known Activations