INDEX

Explanations

terms related to ownership or intellectual property

oai_token-act-pair · gpt-3.5-turbo

terms related to proprietary rights and trademarks

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

sum

-0.74

stones

-0.73

let

-0.68

saw

-0.68

once

-0.65

lad

-0.64

ont

-0.63

hat

-0.63

eating

-0.62

esh

-0.61

POSITIVE LOGITS

 proprietary

3.84

 patented

1.82

 confidential

1.21

 patent

1.18

 copyrighted

1.18

 patents

1.17

 commercially

1.09

 customized

1.05

 standardized

1.03

 subsidiary

1.02

Activations Density 0.023%

terms related to ownership or intellectual property

terms related to proprietary rights and trademarks

No Comments

No Known Activations