INDEX

Explanations

the word "par" followed by a number indicating strong activation

oai_token-act-pair · gpt-3.5-turbo

instances of the word "par" in various 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

ħĭ

-0.70

Accessory

-0.70

sonian

-0.64

 Territories

-0.63

OME

-0.61

¥µ

-0.60

 sockets

-0.60

 Tablet

-0.58

 discovery

-0.57

hirt

-0.57

POSITIVE LOGITS

rots

1.12

rot

1.07

abolic

1.01

agon

0.98

anoia

0.98

odies

0.98

leys

0.97

iah

0.97

odic

0.95

rish

0.95

Activations Density 0.015%

the word "par" followed by a number indicating strong activation

instances of the word "par" in various contexts

No Comments

No Known Activations