INDEX

Explanations

autographed items or activities related to autographing

oai_token-act-pair · gpt-3.5-turbo

words related to autographs and their context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

dip

-0.66

 exits

-0.65

Priv

-0.64

ava

-0.63

inn

-0.62

Asset

-0.61

Ve

-0.61

-+

-0.60

 fork

-0.59

 travellers

-0.59

POSITIVE LOGITS

ographed

3.77

kus

1.45

ockey

1.35

ograp

1.34

cca

1.07

okin

1.02

 Hiroshima

1.01

 Reign

0.90

oland

0.85

olitics

0.84

Activations Density 0.033%

autographed items or activities related to autographing

words related to autographs and their context

No Comments

No Known Activations