INDEX

Explanations

names of people, places, or things, especially "Troy"

oai_token-act-pair · gpt-3.5-turbo

occurrences of the name "Troy" 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 @ 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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

FML

-1.03

chi

-0.78

SPA

-0.74

scl

-0.74

 exercise

-0.73

 encl

-0.73

elf

-0.72

PLA

-0.71

hani

-0.68

 lift

-0.68

POSITIVE LOGITS

 Troy

3.88

 Tulsa

1.39

 Syracuse

1.29

roy

1.24

Tul

1.19

 Hector

1.18

 Homer

1.15

 Albany

1.15

 Kent

1.11

 Auburn

1.10

Activations Density 0.013%

names of people, places, or things, especially "Troy"

occurrences of the name "Troy" in various contexts

No Comments

No Known Activations