INDEX

Explanations

extracted numbers and quantities from the text

oai_token-act-pair · gpt-3.5-turbo

punctuation marks, specifically commas

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 maiden

-0.66

wom

-0.65

 sights

-0.65

 enthusi

-0.57

 heights

-0.55

 phys

-0.55

Ely

-0.54

 affairs

-0.52

 stout

-0.52

 visuals

-0.50

POSITIVE LOGITS

1.72

1.67

1.65

1.64

1.59

1.58

1.49

1.43

Activations Density 0.058%

extracted numbers and quantities from the text

punctuation marks, specifically commas

No Comments

No Known Activations