INDEX

Explanations

social media related information such as blocked accounts, following, and dates

oai_token-act-pair · gpt-3.5-turbo

specific words or phrases related to being blocked or canceled in social media or similar contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 scout

-0.85

theless

-0.78

viz

-0.72

 dart

-0.72

 bolt

-0.70

cu

-0.69

 coupled

-0.69

dj

-0.67

 explorer

-0.67

 paired

-0.66

POSITIVE LOGITS

erity

1.22

rupted

1.08

izophren

1.00

ueless

0.97

rupt

0.96

seless

0.95

untary

0.95

secut

0.94

ificantly

0.94

bitious

0.93

Activations Density 0.256%

social media related information such as blocked accounts, following, and dates

specific words or phrases related to being blocked or canceled in social media or similar contexts

No Comments

No Known Activations