INDEX

Explanations

phrases related to news reporting and requests for comment

oai_token-act-pair · gpt-3.5-turbo

instances of requests for comments or responses in a text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 unstoppable

-0.87

 toler

-0.76

 lifes

-0.76

 awakening

-0.75

 exploding

-0.74

 amazing

-0.74

 thriving

-0.74

 insane

-0.73

 sucker

-0.73

 tremend

-0.71

POSITIVE LOGITS

 However

1.09

 Asked

1.08

<|endoftext|>

1.04

 Emails

1.03

 Nevertheless

1.03

 Sources

1.02

 Nonetheless

1.02

 Additionally

1.01

 Officials

1.01

 Â®

1.01

Activations Density 0.262%

phrases related to news reporting and requests for comment

instances of requests for comments or responses in a text

No Comments

No Known Activations