INDEX

Explanations

unique or special items referred to as "Original" or "original."

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of the term "Original" 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 @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

wal

-0.85

ega

-0.71

rosso

-0.71

roph

-0.69

ostics

-0.68

ucket

-0.67

robe

-0.67

rolet

-0.66

rology

-0.66

walk

-0.65

POSITIVE LOGITS

ity

1.42

ITY

0.92

 incarnation

0.88

sin

0.86

 impetus

0.86

 intent

0.83

 conception

0.81

 trilogy

0.81

 screenplay

0.80

 intention

0.79

Activations Density 0.028%

unique or special items referred to as "Original" or "original."

repeated mentions of the term "Original" in various contexts

No Comments

No Known Activations