INDEX

Explanations

specific names and quotes

oai_token-act-pair · gpt-3.5-turbo

quotations and dialogue from individuals

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

vec

-0.95

 asbestos

-0.89

bestos

-0.86

vec

-0.84

 stim

-0.82

acco

-0.81

 âĸĪ

-0.81

 respir

-0.80

 neuroscience

-0.78

 flashlight

-0.76

POSITIVE LOGITS

King

1.99

 King

1.94

 Kingdom

1.69

 KING

1.67

 kingdom

1.67

 monarch

1.64

 kingdoms

1.64

 king

1.63

 Kings

1.61

 kings

1.56

Activations Density 0.354%

specific names and quotes

quotations and dialogue from individuals

No Comments

No Known Activations