INDEX

Explanations

phrases related to methods, procedures, or directions

oai_token-act-pair · gpt-3.5-turbo

references to "ways" or methods of doing something

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

izen

-0.74

RANT

-0.68

bach

-0.67

MSN

-0.61

oler

-0.60

 Edited

-0.60

images

-0.60

sen

-0.59

bers

-0.59

illon

-0.59

POSITIVE LOGITS

way

3.55

WAY

2.56

Way

2.12

way

1.92

 ways

1.85

 manner

1.76

Way

1.49

WAY

1.45

ways

1.27

 Ways

1.09

Activations Density 0.065%

phrases related to methods, procedures, or directions

references to "ways" or methods of doing something

No Comments

No Known Activations