INDEX

Explanations

the word "Prot" followed by a number in the context of discussing different types of prototypes

oai_token-act-pair · gpt-3.5-turbo

terms related to prototypes and protection

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

manship

-0.90

BRE

-0.74

 Confederacy

-0.61

 Crusher

-0.61

 McGee

-0.60

Gap

-0.59

 clust

-0.59

makers

-0.57

 Kinn

-0.57

 DRAG

-0.57

POSITIVE LOGITS

ected

1.64

otyp

1.52

otypes

1.51

racted

1.44

agonist

1.44

ractor

1.42

ective

1.39

ector

1.39

ocol

1.36

ests

1.36

Activations Density 0.037%

the word "Prot" followed by a number in the context of discussing different types of prototypes

terms related to prototypes and protection

No Comments

No Known Activations