INDEX

Explanations

linguistic patterns where a specific word or phrase is embedded within other words

oai_token-act-pair · gpt-3.5-turbo

references to code structure and inline elements, particularly focusing on "inline" usage and linemen in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 civilisation

-0.78

ctuary

-0.78

 Lodge

-0.78

Jem

-0.76

Moon

-0.75

Hab

-0.74

Sac

-0.73

 Bapt

-0.73

Gy

-0.73

eem

-0.73

POSITIVE LOGITS

 inline

3.03

inline

2.17

 lineman

1.82

Xin

1.53

 linemen

1.52

Async

1.35

IL

1.04

IPv

0.98

 async

0.98

std

0.92

Activations Density 0.048%

linguistic patterns where a specific word or phrase is embedded within other words

references to code structure and inline elements, particularly focusing on "inline" usage and linemen in various contexts

No Comments

No Known Activations