INDEX

Explanations

locations frequented by people and objects

oai_token-act-pair · gpt-3.5-turbo

references to locations and types of establishments

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

maxwell

-0.68

CHAT

-0.65

 appetite

-0.64

PUT

-0.64

anan

-0.63

XT

-0.62

ILA

-0.61

onite

-0.59

ilogy

-0.58

 Rebellion

-0.58

POSITIVE LOGITS

hops

1.27

chool

1.22

paces

1.21

hips

1.20

 alike

1.07

 nationwide

1.05

 frequ

1.05

 across

1.03

afety

0.99

 throughout

0.98

Activations Density 0.235%

locations frequented by people and objects

references to locations and types of establishments

No Comments

No Known Activations