INDEX

Explanations

phrases relating to the application or use of concepts or rules

oai_token-act-pair · gpt-3.5-turbo

instances of the word "apply" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ument

-0.68

footed

-0.66

airs

-0.63

rous

-0.62

omes

-0.62

ifa

-0.60

OUS

-0.60

 Trog

-0.60

acters

-0.60

ymes

-0.60

POSITIVE LOGITS

 inconsist

0.72

 pressure

0.71

enz

0.71

aution

0.71

 uniformly

0.69

rences

0.67

 sunscreen

0.66

 Anarchy

0.66

rals

0.66

 brakes

0.64

Activations Density 0.035%

phrases relating to the application or use of concepts or rules

instances of the word "apply" in various contexts

No Comments

No Known Activations