INDEX

Explanations

mentions of specific words like 'Stroke', 'Prince', and 'Principle'

oai_token-act-pair · gpt-3.5-turbo

references to the term "Stroke" and its various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

iom

-0.76

idity

-0.69

igraph

-0.66

igrant

-0.66

ICAN

-0.65

ãģĤ

-0.65

 duck

-0.65

 SCHOOL

-0.64

<-

-0.62

boat

-0.62

POSITIVE LOGITS

 Stro

0.89

 exha

0.70

 specificity

0.70

 Mayweather

0.68

ascus

0.67

 Subcommittee

0.66

 Examination

0.65

 Participation

0.65

 frust

0.64

phabet

0.64

Activations Density 0.001%

mentions of specific words like 'Stroke', 'Prince', and 'Principle'

references to the term "Stroke" and its various contexts

No Comments

No Known Activations