INDEX

Explanations

names of people related to legal or political contexts

oai_token-act-pair · gpt-3.5-turbo

mentions of individuals and entities related to the media, particularly Gawker and Hogan

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

igmatic

-0.86

rics

-0.84

rica

-0.78

¯

-0.76

 Dying

-0.72

illing

-0.67

ql

-0.67

ks

-0.64

omething

-0.63

ighter

-0.63

POSITIVE LOGITS

ulkan

0.88

 Hogan

0.88

 Ukrain

0.78

 amphib

0.65

Ic

0.63

 Amph

0.63

rieved

0.62

Mania

0.62

 Yanukovych

0.62

 prefrontal

0.61

Activations Density 0.040%

names of people related to legal or political contexts

mentions of individuals and entities related to the media, particularly Gawker and Hogan

No Comments

No Known Activations