INDEX

Explanations

phrases related to making or not making requests

oai_token-act-pair · gpt-3.5-turbo

references to requests for comments or information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Sul

-0.69

 Brill

-0.67

nat

-0.67

 Blacks

-0.65

 Chop

-0.65

noxious

-0.65

 Brilliant

-0.63

otine

-0.62

tal

-0.62

 Barkley

-0.61

POSITIVE LOGITS

 requests

1.16

 request

1.04

 permission

0.96

request

0.92

 warrants

0.90

 requested

0.87

ioned

0.86

aires

0.85

 requesting

0.83

naires

0.81

Activations Density 0.024%

phrases related to making or not making requests

references to requests for comments or information

No Comments

No Known Activations