INDEX

Explanations

phrases related to technology, especially mentions of specific software or gadgets

oai_token-act-pair · gpt-3.5-turbo

references to a specific brand or product

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

 Lutheran

-0.81

Fram

-0.67

 lesson

-0.65

 Lisbon

-0.64

 Copenhagen

-0.63

 Wand

-0.63

 Kits

-0.62

piece

-0.61

ĺħ

-0.61

 Shack

-0.61

POSITIVE LOGITS

oug

1.61

hest

1.07

etooth

0.92

Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯

0.89

anium

0.84

terness

0.83

heed

0.81

hesda

0.81

ileaks

0.81

vre

0.80

Activations Density 0.005%

phrases related to technology, especially mentions of specific software or gadgets

references to a specific brand or product

No Comments

No Known Activations