INDEX

Explanations

locations or places

oai_token-act-pair · gpt-3.5-turbo

punctuation marks, primarily periods

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 endeav

-0.79

 neglect

-0.79

 mathemat

-0.78

 unexplained

-0.78

 plaus

-0.77

 persuasion

-0.77

 bending

-0.75

 modelling

-0.75

 stubborn

-0.74

 inertia

-0.74

POSITIVE LOGITS

 Additionally

1.37

Tickets

1.34

 Tickets

1.19

Additionally

1.12

<|endoftext|>

1.08

 Additional

1.07

 NOTE

1.06

 Previously

1.02

 Also

1.02

 Fans

1.00

Activations Density 0.304%

locations or places

punctuation marks, primarily periods

No Comments

No Known Activations