INDEX

Explanations

specific product descriptions related to various items

oai_token-act-pair · gpt-3.5-turbo

references to products or items being described positively

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

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

Jews

-0.88

rollment

-0.86

redits

-0.85

termination

-0.81

aucuses

-0.79

igration

-0.79

Muslims

-0.77

Players

-0.76

Dialogue

-0.76

ugu

-0.76

POSITIVE LOGITS

 versatile

1.15

 product

1.13

 inexpensive

1.11

 thing

1.11

kit

1.09

 sturdy

1.09

 item

1.06

 homemade

1.01

 sleek

1.01

 sucker

0.99

Activations Density 0.219%

specific product descriptions related to various items

references to products or items being described positively

No Comments

No Known Activations