INDEX

Explanations

names of people

oai_token-act-pair · gpt-3.5-turbo

names of notable athletes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Rowling

-0.83

 Gloria

-0.71

 Feminist

-0.71

 Lanka

-0.68

Wem

-0.68

 Canaver

-0.67

 censor

-0.67

 Bulg

-0.66

 mathematic

-0.64

 Squid

-0.64

POSITIVE LOGITS

Jr

1.24

III

1.00

 teammate

0.90

 rookie

0.90

 injured

0.89

 injury

0.89

 Rookie

0.87

 vacated

0.84

 Injury

0.84

Sr

0.84

Activations Density 0.180%

names of people

names of notable athletes

No Comments

No Known Activations