INDEX

Explanations

instances of the word "bit" or similar descriptive terms

oai_token-act-pair · gpt-3.5-turbo

references to quantities, specifically variations of "bit" and "lot."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

urated

-0.76

aring

-0.71

cession

-0.68

ym

-0.66

drawn

-0.64

 Registration

-0.60

hold

-0.60

eton

-0.58

 Ragnarok

-0.58

VK

-0.58

POSITIVE LOGITS

 differently

0.71

 anyways

0.70

 afterward

0.69

;)

0.69

 because

0.66

:-)

0.64

:)

0.64

ago

0.64

bc

0.64

 lately

0.64

Activations Density 0.238%

instances of the word "bit" or similar descriptive terms

references to quantities, specifically variations of "bit" and "lot."

No Comments

No Known Activations