INDEX

Explanations

phrases indicating something being sent back or returned to a previous state

oai_token-act-pair · gpt-3.5-turbo

suffixes related to modification or change

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

 prof

-0.63

 Macedonia

-0.63

 simulator

-0.62

 Judah

-0.62

 conjecture

-0.61

cal

-0.61

 Vanity

-0.61

 tears

-0.61

 bulls

-0.61

proc

-0.60

POSITIVE LOGITS

anding

4.17

anded

3.82

ands

2.11

ansion

1.56

anders

1.25

ander

1.19

onential

1.19

and

1.18

AND

1.17

ando

1.16

Activations Density 0.007%

phrases indicating something being sent back or returned to a previous state

suffixes related to modification or change

No Comments

No Known Activations