INDEX

Explanations

words ending in "person" or "people" in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to individuals or personas in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Triangle

-0.71

é¾

-0.68

DERR

-0.67

 Monaco

-0.66

æĸ¹

-0.62

Bav

-0.61

 Franken

-0.61

 Tycoon

-0.61

 minus

-0.61

Rated

-0.61

POSITIVE LOGITS

pect

1.11

hip

1.01

pers

0.99

istence

0.98

erver

0.95

istent

0.85

cient

0.83

afety

0.82

ervative

0.82

icker

0.80

Activations Density 0.004%

words ending in "person" or "people" in various contexts

references to individuals or personas in various contexts

No Comments

No Known Activations