INDEX

Explanations

words related to the concept of being locked or stuck

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "lock."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

oun

-0.75

haps

-0.70

issan

-0.67

PsyNetMessage

-0.67

OTAL

-0.67

¶æ

-0.66

meric

-0.66

sembly

-0.63

Pwr

-0.62

alsa

-0.61

POSITIVE LOGITS

heed

1.01

picking

0.97

hart

0.95

lear

0.94

ingly

0.81

ament

0.81

er

0.81

ERY

0.79

strap

0.77

gren

0.77

Activations Density 0.015%

words related to the concept of being locked or stuck

occurrences of the word "lock."

No Comments

No Known Activations