INDEX

Explanations

mentions of the word "Trotsky."

oai_token-act-pair · gpt-3.5-turbo

references to "Tro" followed by a number, indicating a focus on individuals or characters related to a specific context or universe

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

,,,,

-0.72

icago

-0.70

 INTER

-0.66

à¼

-0.66

-0.65

soever

-0.64

ITNESS

-0.64

ieval

-0.63

 Reloaded

-0.62

 USPS

-0.60

POSITIVE LOGITS

Tro

1.25

tro

1.05

dden

1.04

phies

0.98

opers

0.94

opa

0.88

kefeller

0.87

plets

0.84

ttle

0.83

pha

0.81

Activations Density 0.007%

mentions of the word "Trotsky."

references to "Tro" followed by a number, indicating a focus on individuals or characters related to a specific context or universe

No Comments

No Known Activations