INDEX

Explanations

forms or types of concepts or objects

oai_token-act-pair · gpt-3.5-turbo

terms related to types and forms of concepts or classifications

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

 DRAG

-0.61

 Doors

-0.60

 attachments

-0.59

paces

-0.58

 Bars

-0.58

 Robots

-0.57

 IMAGES

-0.57

 submissions

-0.56

ources

-0.56

reports

-0.55

POSITIVE LOGITS

less

0.89

of

0.86

istically

0.85

ically

0.84

atical

0.84

ally

0.82

ially

0.81

ificant

0.79

osite

0.76

ier

0.76

Activations Density 0.220%

forms or types of concepts or objects

terms related to types and forms of concepts or classifications

No Comments

No Known Activations