INDEX

Explanations

phrases indicating a negative event or action associated with individuals or groups

oai_token-act-pair · gpt-3.5-turbo

the occurrence of the verb "were."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Flavoring

-0.59

emis

-0.55

Merit

-0.53

 Seller

-0.51

position

-0.50

itionally

-0.50

ËĪ

-0.49

Effective

-0.49

ebin

-0.48

POSITIVE LOGITS

 were

2.81

 weren

2.50

were

2.49

 Were

1.97

Were

1.90

are

1.69

 aren

1.48

're

1.39

 hadn

1.32

 differed

1.24

Activations Density 0.169%

phrases indicating a negative event or action associated with individuals or groups

the occurrence of the verb "were."

No Comments

No Known Activations