INDEX

Explanations

phrases related to specific actions and discussions in a community or society

oai_token-act-pair · gpt-3.5-turbo

various forms of numerical or statistical references

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 emperor

-0.83

 derog

-0.81

 traitor

-0.81

 treason

-0.79

 deliber

-0.78

 assassin

-0.78

 aven

-0.78

 reflex

-0.77

 consum

-0.77

 occult

-0.76

POSITIVE LOGITS

But

1.46

Residents

1.46

Instead

1.42

Copyright

1.39

Some

1.39

They

1.37

Currently

1.36

Because

1.36

Advertisement

1.36

Owner

1.36

Activations Density 0.486%

phrases related to specific actions and discussions in a community or society

various forms of numerical or statistical references

No Comments

No Known Activations