INDEX

Explanations

editor's notes in a text

oai_token-act-pair · gpt-3.5-turbo

possessive forms, particularly indicating ownership or attribution 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 @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

yz

-0.80

arians

-0.74

icity

-0.72

ãģ®éŃĶ

-0.70

agara

-0.70

ĪĴ

-0.70

enn

-0.69

facts

-0.69

lain

-0.69

oken

-0.68

POSITIVE LOGITS

 Picks

1.28

 note

1.16

 Note

1.16

 Guide

0.95

 Notes

0.94

 Guild

0.93

 Choice

0.92

 Recommend

0.91

 remorse

0.90

 Commentary

0.86

Activations Density 0.048%

editor's notes in a text

possessive forms, particularly indicating ownership or attribution in a text

No Comments

No Known Activations