INDEX

Explanations

the name "Correa" in different contexts

oai_token-act-pair · gpt-3.5-turbo

mentioning of the name "Correa."

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 icon

-0.72

 charts

-0.68

 icons

-0.67

 skill

-0.64

 downt

-0.63

 goodwill

-0.63

 chart

-0.62

 disabling

-0.62

 physical

-0.62

 marked

-0.61

POSITIVE LOGITS

rea

4.68

romeda

1.72

rett

1.22

resa

1.18

rez

1.15

re

1.13

reon

1.12

rÃ©

1.12

rine

1.10

 Andrea

1.10

Activations Density 0.017%

the name "Correa" in different contexts

mentioning of the name "Correa."

No Comments

No Known Activations