INDEX

Explanations

instances where a change or action is being described as happening or being done

oai_token-act-pair · gpt-3.5-turbo

statements indicating consequences or implications

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

 Saud

-0.74

inav

-0.69

adh

-0.66

ustration

-0.61

¿½

-0.57

 Abdullah

-0.57

ovo

-0.57

 condol

-0.57

 Hyder

-0.56

holm

-0.56

POSITIVE LOGITS

pora

0.72

 automatically

0.71

 usable

0.68

 liable

0.64

 unus

0.62

 fewer

0.62

bered

0.62

literally

0.61

forth

0.61

 theoretically

0.61

Activations Density 0.358%

instances where a change or action is being described as happening or being done

statements indicating consequences or implications

No Comments

No Known Activations