INDEX

Explanations

phrases related to physical interactions, control, or authority

oai_token-act-pair · gpt-3.5-turbo

phrases related to power dynamics and hierarchical positions

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

ratulations

-0.75

ortium

-0.66

phis

-0.66

avorite

-0.65

sylv

-0.65

eur

-0.63

theless

-0.59

 spores

-0.59

 valuable

-0.58

izoph

-0.57

POSITIVE LOGITS

 behest

1.06

 helm

0.86

 altar

0.83

 expense

0.80

 outset

0.80

 urging

0.79

 discretion

0.79

 table

0.78

 periphery

0.77

 mercy

0.73

Activations Density 0.123%

phrases related to physical interactions, control, or authority

phrases related to power dynamics and hierarchical positions

No Comments

No Known Activations