INDEX

Explanations

the letter "P" followed by a number

oai_token-act-pair · gpt-3.5-turbo

references to a specific variable or parameter labeled as 'P'

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 diplom

-0.65

 Lauder

-0.64

 Jama

-0.63

 neoc

-0.63

 orche

-0.61

Rih

-0.61

 ancest

-0.60

 kinderg

-0.60

 Glas

-0.59

 vested

-0.59

POSITIVE LOGITS

airs

1.28

adding

1.24

ixels

1.24

ivot

1.20

aired

1.18

ainted

1.15

EEK

1.15

ipes

1.14

icking

1.13

ylon

1.12

Activations Density 0.040%

the letter "P" followed by a number

references to a specific variable or parameter labeled as 'P'

No Comments

No Known Activations