INDEX

Explanations

phrases related to entering or beginning involvement in a specific activity or subject

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate involvement or engagement in activities

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

erence

-0.69

 bene

-0.62

\'

-0.60

agues

-0.60

lessly

-0.60

âĤ¬

-0.60

selves

-0.60

ankind

-0.59

men

-0.59

 Glob

-0.59

POSITIVE LOGITS

 trouble

1.12

 mischief

0.81

itialized

0.79

 puberty

0.79

 politics

0.78

 grips

0.78

 hypers

0.77

 shape

0.77

bed

0.76

EStream

0.75

Activations Density 0.054%

phrases related to entering or beginning involvement in a specific activity or subject

phrases that indicate involvement or engagement in activities

No Known Activations