INDEX

Explanations

proper nouns or specific names of programs, bills, laws, or initiatives

oai_token-act-pair · gpt-3.5-turbo

phrases that reference specific laws or bills

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

abuse

-0.76

 texts

-0.72

rooms

-0.71

 qualifies

-0.71

issues

-0.69

 tendencies

-0.69

 newspapers

-0.68

imposed

-0.68

 councillors

-0.68

SPONSORED

-0.68

POSITIVE LOGITS

 Butterfly

0.97

 Excellence

0.94

 Friendship

0.94

 Million

0.93

 Integrated

0.91

 Colleg

0.90

 Employee

0.90

 Courage

0.90

 Gauntlet

0.89

 Responsibility

0.88

Activations Density 0.265%

proper nouns or specific names of programs, bills, laws, or initiatives

phrases that reference specific laws or bills

No Comments

No Known Activations