INDEX

Explanations

names or words ending with "ter"

oai_token-act-pair · gpt-3.5-turbo

occurrences of specific characters or sequences of letters

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

 tremend

-0.88

#$#$

-0.80

 anonymity

-0.77

omaly

-0.75

 privacy

-0.74

ĨĴ

-0.72

 motives

-0.68

 boredom

-0.67

 accuracy

-0.67

ĺħ

-0.67

POSITIVE LOGITS

inki

0.79

ails

0.75

akeru

0.75

qt

0.73

abo

0.72

ciating

0.72

ony

0.71

erella

0.69

asma

0.69

obl

0.69

Activations Density 0.111%

names or words ending with "ter"

occurrences of specific characters or sequences of letters

No Comments

No Known Activations