INDEX

Explanations

verbs related to actions and events

oai_token-act-pair · gpt-3.5-turbo

references to actions and outcomes related to missing or being charged with an offense

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 dont

-0.68

 didnt

-0.67

Newsletter

-0.63

âĢ¦)

-0.63

 Awakens

-0.62

ffen

-0.62

Sly

-0.62

ses

-0.59

 stumble

-0.57

)|

-0.57

POSITIVE LOGITS

etheless

0.82

senal

0.68

asper

0.63

ãĤ´

0.63

ģĸ

0.62

å®

0.62

helial

0.62

olin

0.61

ogether

0.60

addon

0.60

Activations Density 1.223%

verbs related to actions and events

references to actions and outcomes related to missing or being charged with an offense

No Comments

No Known Activations