INDEX

Explanations

references to the word "pent" with varying levels of relevance

oai_token-act-pair · gpt-3.5-turbo

references to the term "pent" in various contexts, likely related to pentagrams or penthouses

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ãģĦ

-0.82

 Shake

-0.71

 Mahar

-0.65

ï¸ı

-0.63

 Ready

-0.61

 advances

-0.58

 Clone

-0.57

 Combine

-0.56

 Gaga

-0.54

 flourish

-0.54

POSITIVE LOGITS

agram

1.40

agonal

1.39

agon

1.34

agons

1.31

athlon

1.16

acles

1.15

acle

1.11

obar

1.09

house

1.09

othal

1.08

Activations Density 0.041%

references to the word "pent" with varying levels of relevance

references to the term "pent" in various contexts, likely related to pentagrams or penthouses

No Comments

No Known Activations