INDEX

Explanations

large numbers (likely represented in a monetary context)

oai_token-act-pair · gpt-3.5-turbo

various numerical data points or statistics

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

 fault

-0.67

 biography

-0.62

ucci

-0.61

 organized

-0.61

 joke

-0.60

 Sheet

-0.60

 blame

-0.60

Ã±a

-0.59

 jokes

-0.58

 vested

-0.58

POSITIVE LOGITS

3.14

2.22

2.21

2.06

2.05

1.96

1.93

1.92

1.88

1.87

Activations Density 0.024%

large numbers (likely represented in a monetary context)

various numerical data points or statistics

No Comments

No Known Activations