INDEX

Explanations

phrases related to downward movement or decrease

oai_token-act-pair · gpt-3.5-turbo

instances of the word "down."

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

 Applied

-0.76

ici

-0.74

alde

-0.67

itizen

-0.67

BLIC

-0.65

formed

-0.63

nor

-0.61

ami

-0.61

heid

-0.59

 TYPE

-0.59

POSITIVE LOGITS

 down

3.18

down

2.47

 Down

2.21

Down

2.13

 DOWN

2.06

 downs

2.03

downs

2.01

DOWN

1.61

 downwards

1.57

 downward

1.37

Activations Density 0.064%

phrases related to downward movement or decrease

instances of the word "down."

No Comments

No Known Activations