INDEX

Explanations

religious or formal language requesting actions or declarations

oai_token-act-pair · gpt-3.5-turbo

imperative verbs and expressions that request action or attention

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

artney

-0.87

FU

-0.70

pload

-0.63

CLR

-0.62

odder

-0.61

iae

-0.60

fts

-0.60

iscons

-0.58

affer

-0.57

adh

-0.57

POSITIVE LOGITS

 myself

0.89

²¾

0.79

poke

0.75

arest

0.73

 paraph

0.66

xus

0.64

 sincerely

0.64

fully

0.63

Ļ

0.62

 ourselves

0.61

Activations Density 0.168%

religious or formal language requesting actions or declarations

imperative verbs and expressions that request action or attention

No Comments

No Known Activations