INDEX

Explanations

unusual or out of the ordinary occurrences

oai_token-act-pair · gpt-3.5-turbo

instances of the word "strange" and related concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ainers

-0.79

vation

-0.79

ensional

-0.77

dain

-0.77

agles

-0.75

cussion

-0.75

amaru

-0.74

chen

-0.74

payer

-0.74

rien

-0.73

POSITIVE LOGITS

 occurrences

0.98

 coincidence

0.90

worldly

0.89

 anomalies

0.86

ly

0.84

 anomaly

0.82

ness

0.79

 twists

0.79

lihood

0.79

 strange

0.78

Activations Density 0.018%

unusual or out of the ordinary occurrences

instances of the word "strange" and related concepts

No Comments

No Known Activations