INDEX

Explanations

references to turbo-related terms such as 'turbo', 'turbocharge', and 'turbos', with varying activation strengths

oai_token-act-pair · gpt-3.5-turbo

terms related to turbocharging systems or enhancements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

soever

-0.92

*/(

-0.85

ä»£

-0.77

 Flavoring

-0.74

Ö¼

-0.72

 Debor

-0.71

arians

-0.71

 Correspond

-0.71

 Marketable

-0.70

utical

-0.70

POSITIVE LOGITS

charg

1.35

charged

1.31

charging

1.29

charges

0.96

charge

0.93

 absor

0.90

 turbo

0.89

plug

0.88

Gra

0.86

jet

0.85

Activations Density 0.006%

references to turbo-related terms such as 'turbo', 'turbocharge', and 'turbos', with varying activation strengths

terms related to turbocharging systems or enhancements

No Comments

No Known Activations