INDEX

Explanations

large numeric values

oai_token-act-pair · gpt-3.5-turbo

numerical values and their associated units in a context related to property sizes

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 ends

-0.71

 Gillespie

-0.70

 susp

-0.67

 Minerva

-0.66

 Nasa

-0.65

 Clifford

-0.65

 Instruct

-0.65

 exceptions

-0.64

Cox

-0.63

 reperc

-0.62

POSITIVE LOGITS

pound

1.43

plus

1.38

dollar

1.37

capacity

1.37

strong

1.32

mile

1.32

year

1.27

page

1.21

member

1.20

veh

1.19

Activations Density 0.033%

large numeric values

numerical values and their associated units in a context related to property sizes

No Comments

No Known Activations