INDEX

Explanations

words related to creating or updating content

oai_token-act-pair · gpt-3.5-turbo

instances of creation and publication of content

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

backs

-0.79

bill

-0.76

stood

-0.76

adra

-0.75

raf

-0.72

adish

-0.71

bull

-0.71

BL

-0.70

nder

-0.70

vor

-0.70

POSITIVE LOGITS

 Created

1.15

ablishment

0.93

 Rating

0.80

CLASSIFIED

0.80

 Detected

0.79

 Thumbnails

0.77

Created

0.77

 Coverage

0.75

 Updated

0.75

>[

0.73

Activations Density 0.007%

words related to creating or updating content

instances of creation and publication of content

No Comments

No Known Activations