INDEX

Explanations

forms of the word "objects" in the text

oai_token-act-pair · gpt-3.5-turbo

references to various types of objects

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

 Hayes

-0.71

starter

-0.70

hhhh

-0.68

illin

-0.67

reon

-0.67

Bay

-0.66

proven

-0.65

season

-0.64

 Hust

-0.64

 mell

-0.64

POSITIVE LOGITS

 objects

3.78

 Objects

2.99

objects

2.91

 object

2.65

object

1.95

Object

1.83

 Object

1.80

 artifacts

1.53

 items

1.52

 instruments

1.38

Activations Density 0.024%

forms of the word "objects" in the text

references to various types of objects

No Comments

No Known Activations