INDEX

Explanations

instances of words related to official documents or records

oai_token-act-pair · gpt-3.5-turbo

references to documentation and its importance in various contexts

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

joined

-0.68

 tempt

-0.68

¦

-0.68

corn

-0.67

leon

-0.67

billion

-0.65

star

-0.64

isson

-0.64

bet

-0.63

opoly

-0.62

POSITIVE LOGITS

 documentation

3.53

 Documentation

2.58

 docs

2.27

 documents

1.73

 paperwork

1.68

 documenting

1.63

 manuals

1.60

Document

1.59

 document

1.52

 documented

1.41

Activations Density 0.014%

instances of words related to official documents or records

references to documentation and its importance in various contexts

No Comments

No Known Activations