INDEX

Explanations

Interview-style dialogue punctuation

oai_token-act-pair · gpt-3.5-turbo

instances of dialogue or spoken phrases within the text

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

 conclude

-0.63

 thereafter

-0.63

senal

-0.63

 subsidized

-0.63

 expense

-0.61

 Coffin

-0.60

 expires

-0.60

 Warehouse

-0.60

 fors

-0.59

 territ

-0.59

POSITIVE LOGITS

Correct

0.82

ONSORED

0.79

Bus

0.79

"_

0.77

âĿ

0.77

Ku

0.72

Of

0.72

Hum

0.71

Asked

0.70

Dialogue

0.69

Activations Density 0.423%

Interview-style dialogue punctuation

instances of dialogue or spoken phrases within the text

No Comments

No Known Activations