INDEX

Explanations

subcategories or subdivisions

oai_token-act-pair · gpt-3.5-turbo

references to subs, such as subcategories and subplots

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ILLE

-0.85

 Walker

-0.71

 Muse

-0.69

 Qiao

-0.69

 Zurich

-0.67

 Moody

-0.67

rot

-0.66

 Cardinal

-0.66

 GOODMAN

-0.66

 Lauder

-0.65

POSITIVE LOGITS

missive

1.48

mitter

1.25

verting

1.24

sequent

1.20

species

1.19

sequently

1.19

mits

1.16

culture

1.16

continental

1.16

mers

1.15

Activations Density 0.018%

subcategories or subdivisions

references to subs, such as subcategories and subplots

No Comments

No Known Activations