INDEX

Explanations

names of products or companies combined with some form of launching or introducing

oai_token-act-pair · gpt-3.5-turbo

instances of commas used to separate elements in a list or series

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ances

-0.81

oÄŁan

-0.81

ance

-0.76

rawdownloadcloneembedreportprint

-0.75

xes

-0.74

quist

-0.72

orts

-0.71

yers

-0.70

Previous

-0.69

ÄŁ

-0.67

POSITIVE LOGITS

 namely

1.40

 dubbed

1.13

 aptly

1.05

 called

1.05

 nicknamed

0.99

 albeit

0.97

 Spectre

0.94

 known

0.91

 consisting

0.89

 termed

0.89

Activations Density 0.257%

names of products or companies combined with some form of launching or introducing

instances of commas used to separate elements in a list or series

No Comments

No Known Activations