INDEX

Explanations

phrases related to negative actions or harmful behaviors

oai_token-act-pair · gpt-3.5-turbo

references to actions and their impact on people

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

SPONSORED

-0.82

ãĥ¡

-0.76

NetMessage

-0.67

Org

-0.62

Orb

-0.61

 WRITE

-0.61

 Albania

-0.61

 Orders

-0.59

normal

-0.58

aucas

-0.58

POSITIVE LOGITS

eering

0.70

 appropriately

0.67

 accordingly

0.67

wards

0.66

 Carbuncle

0.65

dit

0.64

livion

0.64

onto

0.62

olesterol

0.62

othal

0.61

Activations Density 0.600%

phrases related to negative actions or harmful behaviors

references to actions and their impact on people

No Comments

No Known Activations