INDEX

Explanations

specific names, especially "Piper" and "Civic"

oai_token-act-pair · gpt-3.5-turbo

references to the term "Piper" and words associated with civic topics

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

roid

-0.78

chnology

-0.70

ooth

-0.69

 blur

-0.67

ement

-0.65

cell

-0.64

eers

-0.64

ted

-0.64

rem

-0.64

stress

-0.63

POSITIVE LOGITS

iper

0.87

orno

0.80

 "$:/

0.79

ãĤº

0.72

NetMessage

0.70

ENT

0.69

itual

0.69

ãĥ´ãĤ¡

0.69

ENCE

0.68

é¾įå

0.67

Activations Density 0.027%

specific names, especially "Piper" and "Civic"

references to the term "Piper" and words associated with civic topics

No Comments

No Known Activations