INDEX

Explanations

instances where a process or action is perceived as completed or at a significant stage of progression

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the completion, availability, or condition of processes and items

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

 presidents

-0.68

çīĪ

-0.66

 backgrounds

-0.65

 descriptions

-0.60

 Cosponsors

-0.60

Tre

-0.59

 summarizes

-0.57

bats

-0.57

ccording

-0.56

Mog

-0.55

POSITIVE LOGITS

 sufficiently

0.73

 imminent

0.67

oldemort

0.66

 enough

0.66

 irreversible

0.66

readable

0.65

ufficient

0.63

 theirs

0.63

orthy

0.63

pai

0.63

Activations Density 0.843%

instances where a process or action is perceived as completed or at a significant stage of progression

phrases indicating the completion, availability, or condition of processes and items

No Comments

No Known Activations