INDEX

Explanations

phrases related to competitive performance or training

oai_token-act-pair · gpt-3.5-turbo

specific sequences of letters or patterns within words

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

aurus

-0.69

ä¸ī

-0.64

aeda

-0.64

anke

-0.63

 Scalia

-0.62

Ĭ±

-0.62

arag

-0.61

 uncont

-0.61

heter

-0.61

 nominate

-0.61

POSITIVE LOGITS

ings

0.93

coat

0.92

hound

0.88

frog

0.85

downs

0.84

knit

0.84

boarding

0.83

board

0.78

breaker

0.78

down

0.77

Activations Density 0.156%

phrases related to competitive performance or training

specific sequences of letters or patterns within words

No Comments

No Known Activations