INDEX

Explanations

phrases indicating uncertainty or contrast in information

oai_token-act-pair · gpt-3.5-turbo

instances of uncertainty or conditional expressions

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

 Skydragon

-0.80

hitting

-0.70

 Travels

-0.68

ands

-0.68

uilding

-0.67

 PLUS

-0.66

resa

-0.65

aturday

-0.63

ateurs

-0.63

earchers

-0.62

POSITIVE LOGITS

 admittedly

1.13

 occasional

0.91

 technically

0.88

 occasionally

0.85

 somewhat

0.83

 nonetheless

0.83

 exceptions

0.82

 varies

0.81

 caveats

0.80

may

0.80

Activations Density 0.255%

phrases indicating uncertainty or contrast in information

instances of uncertainty or conditional expressions

No Comments

No Known Activations