INDEX

Explanations

pieces, objects, or items within a set, such as "timepiece" or "three-piece suit."

oai_token-act-pair · gpt-3.5-turbo

references to artworks or media components referred to as "pieces."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 shelters

-0.66

 databases

-0.60

 Forums

-0.60

 forests

-0.60

 Indra

-0.59

 influence

-0.59

 Apache

-0.58

 homes

-0.57

 forums

-0.57

 libraries

-0.57

POSITIVE LOGITS

piece

4.63

pieces

3.47

 Piece

1.62

 piece

1.50

parts

1.24

tone

1.24

plate

1.19

pole

1.18

 Pieces

1.10

dress

1.09

Activations Density 0.012%

pieces, objects, or items within a set, such as "timepiece" or "three-piece suit."

references to artworks or media components referred to as "pieces."

No Comments

No Known Activations