INDEX

Explanations

dates and ordinal numbers

oai_token-act-pair · gpt-3.5-turbo

date expressions and ordinal indicators

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

andowski

-0.77

orem

-0.76

ateral

-0.66

erman

-0.64

rote

-0.64

henko

-0.60

achable

-0.60

rompt

-0.59

amara

-0.59

enegger

-0.59

POSITIVE LOGITS

 July

1.45

 September

1.43

 October

1.42

 November

1.41

 August

1.41

 February

1.40

 June

1.40

 December

1.38

 January

1.35

 April

1.34

Activations Density 0.078%

dates and ordinal numbers

date expressions and ordinal indicators

No Comments

No Known Activations