INDEX

Explanations

mentions of specific and unique items or characteristics

oai_token-act-pair · gpt-3.5-turbo

words or phrases associated with specific distinguishing features or characteristics

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

isen

-0.80

awar

-0.79

aiman

-0.77

ASC

-0.77

agan

-0.76

á½

-0.73

ulu

-0.69

nih

-0.68

inen

-0.68

onest

-0.67

POSITIVE LOGITS

 signature

0.93

 breaker

0.73

ATURE

0.69

 signatures

0.69

 hallmark

0.67

 accomplishment

0.66

zag

0.65

board

0.64

 achievement

0.63

 style

0.63

Activations Density 0.014%

mentions of specific and unique items or characteristics

words or phrases associated with specific distinguishing features or characteristics

No Comments

No Known Activations