INDEX

Explanations

short excerpts or quotes from text

oai_token-act-pair · gpt-3.5-turbo

references to excerpts from documents

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ggle

-0.89

vil

-0.68

thood

-0.68

ceans

-0.64

gaard

-0.63

ggles

-0.62

gy

-0.62

kos

-0.62

 drain

-0.62

mos

-0.61

POSITIVE LOGITS

 excerpts

1.30

 excerpt

1.25

cerpt

1.02

 guiActiveUn

1.01

DragonMagazine

0.92

 snippets

0.91

...]

0.89

rompt

0.87

 extracts

0.87

icularly

0.85

Activations Density 0.023%

short excerpts or quotes from text

references to excerpts from documents

No Comments

No Known Activations