INDEX

Explanations

incorrectly spelled words

oai_token-act-pair · gpt-3.5-turbo

references to correctness or accuracy in actions or statements

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

aunt

-0.79

ires

-0.69

hunt

-0.65

assment

-0.65

ike

-0.63

 Rise

-0.63

ofi

-0.59

enture

-0.58

dark

-0.58

our

-0.58

POSITIVE LOGITS

 correctly

3.89

 incorrectly

2.63

 properly

2.37

 accurately

2.25

 wisely

1.90

 wrongly

1.85

 appropriately

1.83

 errone

1.80

 rightly

1.78

 correct

1.69

Activations Density 0.012%

incorrectly spelled words

references to correctness or accuracy in actions or statements

No Comments

No Known Activations