INDEX

Explanations

unwanted or contentious mentions, potentially involving political and legal matters

oai_token-act-pair · gpt-3.5-turbo

phrases indicating significant events or emergencies

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

ĳå£«

-0.76

½

-0.70

ichick

-0.68

²

-0.67

rer

-0.67

ķ

-0.67

ãĥ¼ãĥĨãĤ£

-0.66

Ĥ

-0.65

ãĤ±

-0.64

_.

-0.64

POSITIVE LOGITS

Scientists

0.79

Britain

0.73

 Britain

0.69

BBC

0.69

ccording

0.69

Researchers

0.68

Sorry

0.68

Analysis

0.68

 labour

0.67

 IMAGES

0.67

Activations Density 0.044%

unwanted or contentious mentions, potentially involving political and legal matters

phrases indicating significant events or emergencies

No Comments

No Known Activations