INDEX

Explanations

instances of text where something is specifically labeled or named

oai_token-act-pair · gpt-3.5-turbo

references to labeling and classifications

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

 Aston

-0.81

 practise

-0.69

 Vers

-0.63

nesses

-0.63

 Rolls

-0.62

anta

-0.62

own

-0.62

 programme

-0.61

letico

-0.61

 Conc

-0.61

POSITIVE LOGITS

 labeled

3.61

 labelled

3.13

 labeling

1.89

label

1.77

 tagged

1.70

 categorized

1.66

 labels

1.65

 branded

1.62

 termed

1.57

 label

1.57

Activations Density 0.013%

instances of text where something is specifically labeled or named

references to labeling and classifications

No Comments

No Known Activations