INDEX

Explanations

quantitative values or figures derived from mathematical calculations

oai_token-act-pair · gpt-3.5-turbo

references to calculations and numerical analysis

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

usky

-0.73

boro

-0.68

 featuring

-0.66

orn

-0.65

arna

-0.63

 supportive

-0.63

 Ment

-0.62

Rust

-0.62

angered

-0.62

 Franken

-0.62

POSITIVE LOGITS

 calculations

3.38

 calculation

3.34

 calculating

2.63

 calculate

2.58

 Calcul

2.37

 calculates

2.30

 calculator

2.21

 calcul

1.97

 computation

1.91

 Calculator

1.88

Activations Density 0.022%

quantitative values or figures derived from mathematical calculations

references to calculations and numerical analysis

No Comments

No Known Activations