INDEX

Explanations

references and mentions within texts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 faces

-0.69

 mates

-0.64

ishment

-0.62

uates

-0.61

aways

-0.61

ties

-0.60

 abduction

-0.60

SpaceEngineers

-0.60

ÃŃs

-0.60

ished

-0.60

POSITIVE LOGITS

ource

1.31

heet

1.23

peed

1.18

hift

1.17

ystem

1.17

":["

1.12

pace

1.11

cript

1.10

":[{"

1.05

ilver

1.00

Activations Density 0.153%

references and mentions within texts

No Comments

No Known Activations