INDEX

Explanations

adjectives or phrases that express an evaluation or judgment

oai_token-act-pair · gpt-3.5-turbo

statements that assert characteristics or qualities, often using the phrase "is" to indicate descriptions or evaluations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

inders

-0.76

rike

-0.70

 Chero

-0.66

ummies

-0.65

gg

-0.63

ffe

-0.63

nikov

-0.62

cess

-0.58

arty

-0.58

hero

-0.57

POSITIVE LOGITS

 incidentally

0.85

 presumably

0.83

 admittedly

0.73

 ironically

0.71

ometimes

0.70

PK

0.69

baugh

0.69

 coincided

0.66

?)

0.65

 culminated

0.65

Activations Density 0.358%

adjectives or phrases that express an evaluation or judgment

statements that assert characteristics or qualities, often using the phrase "is" to indicate descriptions or evaluations

No Comments

No Known Activations