INDEX

Explanations

names of specific individuals

oai_token-act-pair · gpt-3.5-turbo

proper nouns, particularly names and surnames

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 corrective

-0.68

 hikers

-0.68

Op

-0.65

ocular

-0.65

ilation

-0.64

appropriately

-0.63

spect

-0.62

UV

-0.62

EMP

-0.62

CL

-0.61

POSITIVE LOGITS

theless

1.24

wald

1.09

die

1.05

erick

1.00

ragon

0.97

fried

0.94

sson

0.93

igham

0.91

fred

0.90

ersen

0.89

Activations Density 0.013%

names of specific individuals

proper nouns, particularly names and surnames

No Comments

No Known Activations