INDEX

Explanations

phrases related to asking questions and seeking information

oai_token-act-pair · gpt-3.5-turbo

determinations of capability or potential actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 honoured

-0.62

 Gork

-0.58

 apologised

-0.58

...

-0.57

 Wellington

-0.56

 flavours

-0.56

 recognised

-0.55

 flavour

-0.54

 realised

-0.54

isations

-0.52

POSITIVE LOGITS

âĢ

1.85

Â»

1.61

ãĢ

1.57

âľ

1.51

âĹı

1.49

Â¨

1.48

Â«

1.47

</

1.42

âĺ

1.40

————————

1.38

Activations Density 3.068%

phrases related to asking questions and seeking information

determinations of capability or potential actions

No Comments

No Known Activations