INDEX

Explanations

phrases related to honorific titles and names

oai_token-act-pair · gpt-3.5-turbo

references to academic institutions or entities associated with research

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

yip

-0.91

arcity

-0.80

iazep

-0.68

kinson

-0.67

ndra

-0.66

ursday

-0.64

vity

-0.60

oulos

-0.60

 begg

-0.60

 recollection

-0.59

POSITIVE LOGITS

crew

0.87

ption

0.86

INAL

0.82

Ķ

0.80

cheon

0.77

acle

0.71

othy

0.70

¬

0.68

İ

0.68

 Cortex

0.67

Activations Density 0.070%

phrases related to honorific titles and names

references to academic institutions or entities associated with research

No Comments

No Known Activations