INDEX

Explanations

mentions of the location "Heidelberg."

oai_token-act-pair · gpt-3.5-turbo

references to individuals associated with infidelity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 effect

-0.75

 props

-0.70

 sensation

-0.67

 influence

-0.65

 boards

-0.61

gag

-0.60

 helmets

-0.60

 mantra

-0.60

 shaving

-0.60

 Stories

-0.59

POSITIVE LOGITS

idel

4.59

idelity

1.62

idem

1.31

idal

1.10

iden

1.07

ischer

1.06

ides

1.01

 Fidel

0.98

lic

0.98

adel

0.98

Activations Density 0.017%

mentions of the location "Heidelberg."

references to individuals associated with infidelity

No Comments

No Known Activations