INDEX

Explanations

phrases related to personal identity and characteristics

oai_token-act-pair · gpt-3.5-turbo

the presence of the word "a" and variations related to identity and roles

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

scenes

-0.82

ernels

-0.75

Ö¼

-0.75

breaks

-0.72

ourses

-0.71

 appointments

-0.69

uden

-0.68

views

-0.67

books

-0.67

execute

-0.67

POSITIVE LOGITS

 hypocr

1.09

 spectator

1.05

 member

1.04

 prostitute

0.98

 follower

0.98

 virgin

0.98

 participant

0.98

 citizen

0.98

 believer

0.97

 bystand

0.97

Activations Density 0.139%

phrases related to personal identity and characteristics

the presence of the word "a" and variations related to identity and roles

No Comments

No Known Activations