INDEX

Explanations

dates or years mentioned in a document

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate a chronological context or time reference

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 implanted

-0.67

 lodged

-0.65

 outwe

-0.64

 behav

-0.63

 distur

-0.63

 istg

-0.60

 implants

-0.59

 ends

-0.59

 nailed

-0.59

 nurt

-0.59

POSITIVE LOGITS

ventory

1.43

cluding

1.38

visible

1.32

strument

1.30

vasive

1.28

cluded

1.28

clude

1.26

sect

1.25

spect

1.25

come

1.22

Activations Density 0.118%

dates or years mentioned in a document

phrases that indicate a chronological context or time reference

No Comments

No Known Activations