INDEX

Explanations

occurrences of the word "imprint" in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to publishing terms and processes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

cker

-0.79

ggle

-0.79

ctory

-0.78

gger

-0.73

ppo

-0.72

annis

-0.71

ghazi

-0.71

VEL

-0.70

Downloadha

-0.70

arters

-0.70

POSITIVE LOGITS

 imprint

1.27

ed

0.97

edIn

0.90

iated

0.90

eering

0.89

romeda

0.84

lishing

0.81

 arrang

0.80

eers

0.79

ank

0.78

Activations Density 0.013%

occurrences of the word "imprint" in various contexts

references to publishing terms and processes

No Comments

No Known Activations