INDEX

Explanations

specific formatting elements in structured data, such as user profiles and percentages

oai_token-act-pair · gpt-3.5-turbo

percentage symbols and their related contexts

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

 Dinosaur

-0.62

Amazon

-0.59

 orche

-0.58

Reviewed

-0.58

 coax

-0.57

flies

-0.57

Dak

-0.56

 overseen

-0.55

Jet

-0.55

Rex

-0.53

POSITIVE LOGITS

3.79

(%

2.38

%%

2.07

1.92

"%

1.87

 Percentage

1.68

%"

1.67

 Percent

1.66

 percentage

1.62

%%

1.61

Activations Density 0.011%

specific formatting elements in structured data, such as user profiles and percentages

percentage symbols and their related contexts

No Comments

No Known Activations