INDEX

Explanations

terms related to decisions or choices

oai_token-act-pair · gpt-3.5-turbo

phrases related to making fun of or mocking situations and people

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 horizont

-0.69

Constructed

-0.62

 heights

-0.62

hov

-0.61

 energies

-0.61

phrine

-0.61

 expectancy

-0.58

abouts

-0.56

verages

-0.56

ga

-0.56

POSITIVE LOGITS

 Murd

0.81

sense

0.74

 dent

0.67

çİĭ

0.64

 disappear

0.63

ends

0.63

 noises

0.61

ichick

0.61

elly

0.61

nell

0.60

Activations Density 0.113%

terms related to decisions or choices

phrases related to making fun of or mocking situations and people

No Comments

No Known Activations