INDEX

Explanations

words related to different styles or categories within a specific domain

oai_token-act-pair · gpt-3.5-turbo

references to various styles or types in a given context

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

 Mald

-0.80

×Ļ×

-0.77

 Abraham

-0.70

DA

-0.69

ESA

-0.68

riel

-0.67

 Patri

-0.67

×Ļ

-0.67

 Wake

-0.66

 Clarke

-0.66

POSITIVE LOGITS

styles

1.41

 styles

1.39

heet

1.10

 Styles

1.07

ologies

0.99

 styling

0.87

hops

0.87

 chops

0.86

¥ŀ

0.85

 sensibilities

0.83

Activations Density 0.004%

words related to different styles or categories within a specific domain

references to various styles or types in a given context

No Comments

No Known Activations