INDEX

Explanations

products or services characterized by numerical ranges, such as discounts or versions

oai_token-act-pair · gpt-3.5-turbo

references to product dimensions or specifications

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

 fifty

-0.75

 sixty

-0.72

 seventy

-0.71

 ninety

-0.71

 forty

-0.71

 sixteen

-0.70

 graveyard

-0.67

 nineteen

-0.66

 thirty

-0.66

 resent

-0.63

POSITIVE LOGITS

1.24

1.22

1.21

1.17

Hour

1.17

year

1.14

minute

1.12

1.11

inch

1.10

Activations Density 0.066%

products or services characterized by numerical ranges, such as discounts or versions

references to product dimensions or specifications

No Comments

No Known Activations