INDEX

Explanations

technical terms related to computer execution, such as "executed" in different contexts

oai_token-act-pair · gpt-3.5-turbo

the term "executed" 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

Myth

-0.67

cup

-0.64

 Spectrum

-0.63

anta

-0.63

 worries

-0.63

Premium

-0.63

Riv

-0.63

rain

-0.63

 concern

-0.63

 Patreon

-0.62

POSITIVE LOGITS

 executed

3.89

 execution

2.46

 execute

2.37

 executing

2.36

 executes

2.30

 executions

2.00

 Execution

1.87

 hanged

1.69

exec

1.51

 assassinated

1.51

Activations Density 0.020%

technical terms related to computer execution, such as "executed" in different contexts

the term "executed" in various contexts

No Comments

No Known Activations