INDEX

Explanations

mentions of appearances or occurrences of various entities or events

oai_token-act-pair · gpt-3.5-turbo

instances of the word "appeared."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 knowing

-0.60

 backlog

-0.60

 nuts

-0.59

PM

-0.57

 shit

-0.57

 fucking

-0.56

 logistics

-0.55

stra

-0.54

FY

-0.54

RF

-0.53

POSITIVE LOGITS

 appeared

3.40

 appears

2.21

 appear

2.12

 seemed

1.91

 emerged

1.78

 resembled

1.74

 appearing

1.72

 surfaced

1.63

 Appears

1.61

 sounded

1.57

Activations Density 0.013%

mentions of appearances or occurrences of various entities or events

instances of the word "appeared."

No Comments

No Known Activations