INDEX

Explanations

mentions of individual names, potentially focusing on last names

oai_token-act-pair · gpt-3.5-turbo

names with the prefix "Tro" and variations of it, indicating a focus on specific individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 Rebirth

-0.68

 Samoa

-0.65

Sof

-0.64

 neon

-0.64

 Marcos

-0.61

 Krypt

-0.61

 Hurricane

-0.61

 Mosul

-0.60

 Hurricanes

-0.60

âĶĢâĶĢ

-0.59

POSITIVE LOGITS

eger

1.28

iman

1.24

ijn

1.19

sett

1.16

hl

1.15

cz

1.14

hn

1.14

mberg

1.13

utenberg

1.12

cko

1.12

Activations Density 0.113%

mentions of individual names, potentially focusing on last names

names with the prefix "Tro" and variations of it, indicating a focus on specific individuals

No Comments

No Known Activations