INDEX

Explanations

adjectives expressing intensity or importance

oai_token-act-pair · gpt-3.5-turbo

adjectives and descriptors indicating magnitude or quality

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 warp

-0.63

 Berks

-0.62

 Leilan

-0.62

Wer

-0.60

Ley

-0.60

 Rath

-0.59

 Warp

-0.59

 mans

-0.59

DM

-0.59

 Kore

-0.59

POSITIVE LOGITS

theless

1.38

terday

1.24

tenance

1.13

etheless

1.03

withstanding

0.96

selves

0.93

usterity

0.92

mosp

0.88

acters

0.85

lihood

0.85

Activations Density 0.249%

adjectives expressing intensity or importance

adjectives and descriptors indicating magnitude or quality

No Comments

No Known Activations