INDEX

Explanations

words related to specific or definite conditions or requirements

oai_token-act-pair · gpt-3.5-turbo

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

 thence

-0.86

Adds

-0.77

NAS

-0.73

dl

-0.69

ĺħ

-0.68

ften

-0.65

usk

-0.63

iddle

-0.63

 anew

-0.62

arer

-0.61

POSITIVE LOGITS

ties

1.62

 kinds

1.50

 types

1.41

 aspects

1.13

 wavelengths

1.07

 segments

1.02

iating

1.00

 sections

1.00

 embodiments

0.99

types

0.98

Activations Density 0.041%

words related to specific or definite conditions or requirements

No Comments

No Known Activations