INDEX

Explanations

phrases indicating partitioning or categorization

oai_token-act-pair · gpt-3.5-turbo

instances of division or categorization

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

press

-0.71

WT

-0.71

inem

-0.70

lam

-0.70

gery

-0.69

efe

-0.69

ity

-0.67

mb

-0.66

aya

-0.66

dar

-0.66

POSITIVE LOGITS

 sections

1.16

 categories

1.16

 subsections

1.16

 manageable

1.15

 smaller

1.10

 three

1.10

 phases

1.08

 tiers

1.07

 four

1.07

two

1.07

Activations Density 0.098%

phrases indicating partitioning or categorization

instances of division or categorization

No Comments

No Known Activations