INDEX

Explanations

references to Google-related terms or products

oai_token-act-pair · gpt-3.5-turbo

references to Google and its services

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

theless

-0.76

sein

-0.73

lied

-0.71

 slaughtered

-0.70

 strugg

-0.68

Stars

-0.67

 Seym

-0.67

bil

-0.67

isky

-0.66

 elder

-0.66

POSITIVE LOGITS

 Google

3.51

Google

3.08

 google

2.73

google

2.24

 Gmail

2.00

 Alphabet

1.82

 Yelp

1.70

 Facebook

1.68

 LinkedIn

1.59

 Motorola

1.58

Activations Density 0.017%

references to Google-related terms or products

references to Google and its services

No Comments

No Known Activations