INDEX

Explanations

words related to specific events or actions

oai_token-act-pair · gpt-3.5-turbo

words related to roles, goals, and formal presentations or discussions

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 oneself

-0.78

htaking

-0.73

arnaev

-0.71

glomer

-0.69

ecause

-0.69

aundering

-0.68

tesy

-0.65

aples

-0.63

byter

-0.62

ead

-0.61

POSITIVE LOGITS

mates

0.79

ings

0.79

 mates

0.79

 counterparts

0.76

mate

0.68

naires

0.67

books

0.67

ãĤ´ãĥ³

0.66

notes

0.66

 selections

0.65

Activations Density 0.441%

words related to specific events or actions

words related to roles, goals, and formal presentations or discussions

No Comments

No Known Activations