INDEX

Explanations

commands or instructions

oai_token-act-pair · gpt-3.5-turbo

action-oriented descriptors or requests

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ËĪ

-0.63

"—

-0.53

ocracy

-0.53

="/

-0.53

],"

-0.52

PsyNetMessage

-0.52

Ö

-0.51

")

-0.50

 Shutterstock

-0.49

microsoft

-0.47

POSITIVE LOGITS

cknowled

0.69

 furthermore

0.62

beit

0.62

sequently

0.61

 latter

0.60

anwhile

0.59

ogether

0.57

oret

0.55

rarily

0.54

quartered

0.54

Activations Density 0.759%

commands or instructions

action-oriented descriptors or requests

No Comments

No Known Activations