INDEX

Explanations

strings of characters that may not form coherent words or phrases

oai_token-act-pair · gpt-3.5-turbo

separators, punctuation, and symbols typically used in formatting content

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

 telling

-0.90

 acknowledged

-0.87

 affirm

-0.83

 headlines

-0.83

 responding

-0.83

 increasingly

-0.82

 prompted

-0.82

 threatened

-0.82

 belie

-0.82

 consequences

-0.81

POSITIVE LOGITS

Micro

1.38

Amb

1.34

Mini

1.34

Royal

1.33

Cola

1.31

Sem

1.31

Gall

1.30

Dual

1.30

Double

1.30

Gal

1.30

Activations Density 0.324%

strings of characters that may not form coherent words or phrases

separators, punctuation, and symbols typically used in formatting content

No Comments

No Known Activations