INDEX

Explanations

locations or places

oai_token-act-pair · gpt-3.5-turbo

references to navigating or returning to a previous context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

awa

-0.62

 Monstrous

-0.55

CVE

-0.55

thal

-0.53

 theoret

-0.51

doms

-0.51

mint

-0.50

 bait

-0.50

 disadvant

-0.49

theless

-0.49

POSITIVE LOGITS

thumbnails

0.80

oir

0.71

quez

0.68

 Gallery

0.65

photo

0.63

 Caption

0.62

emark

0.61

)=(

0.61

ornia

0.60

ãĥĦ

0.59

Activations Density 0.019%

locations or places

references to navigating or returning to a previous context

No Comments

No Known Activations