INDEX

Explanations

words related to interactions or connections between different entities

oai_token-act-pair · gpt-3.5-turbo

instances of the prefix "inter," indicating relationships or interactions between concepts

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

aughs

-0.68

awaru

-0.65

 Chaser

-0.59

ongyang

-0.58

 shake

-0.57

 Rush

-0.56

hyde

-0.56

Ale

-0.56

 Citation

-0.56

 punches

-0.56

POSITIVE LOGITS

racial

1.19

lude

1.17

continental

1.17

disciplinary

1.15

locking

1.11

dimensional

1.10

stitial

1.06

loc

1.02

rup

1.00

linked

1.00

Activations Density 0.020%

words related to interactions or connections between different entities

instances of the prefix "inter," indicating relationships or interactions between concepts

No Comments

No Known Activations