INDEX

Explanations

phrases related to codes or symbols denoting specific actions or information

oai_token-act-pair · gpt-3.5-turbo

instances of symbols or special characters in the text

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

orts

-0.76

 precaution

-0.75

 protective

-0.73

 oven

-0.72

 processed

-0.68

egu

-0.68

 doll

-0.67

 nesting

-0.67

 loft

-0.67

umin

-0.67

POSITIVE LOGITS

––

1.08

advertisement

0.95

âĢ¢âĢ¢

0.92

————

0.91

—-

0.87

they

0.86

âĢķ

0.86

feat

0.84

–

0.84

something

0.82

Activations Density 0.041%

phrases related to codes or symbols denoting specific actions or information

instances of symbols or special characters in the text

No Comments

No Known Activations