INDEX

Explanations

relevant or interesting pieces of information within a document

oai_token-act-pair · gpt-3.5-turbo

phrases that denote information, summaries, or insights

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

gp

-0.78

ãĥīãĥ©

-0.73

ardless

-0.67

thood

-0.67

 Delivery

-0.65

gart

-0.65

status

-0.64

eur

-0.63

ttle

-0.63

wait

-0.62

POSITIVE LOGITS

 snippets

1.21

 pertinent

1.08

 excerpts

1.07

 redacted

1.04

 glean

1.02

 therein

1.02

 quotations

1.00

 omitted

0.97

 annotations

0.96

 herein

0.96

Activations Density 0.490%

relevant or interesting pieces of information within a document

phrases that denote information, summaries, or insights

No Comments

No Known Activations