INDEX

Explanations

phrases related to revealing or uncovering something hidden or unknown

oai_token-act-pair · gpt-3.5-turbo

metaphors related to openings and barriers

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

VERTISEMENT

-0.81

Ranked

-0.64

iHUD

-0.62

 Entity

-0.59

ija

-0.58

 Replacement

-0.58

 Genius

-0.58

asus

-0.57

KNOWN

-0.56

Pic

-0.56

POSITIVE LOGITS

 offensively

0.89

 nicely

0.78

oos

0.78

 forcefully

0.78

 awhile

0.77

 diplom

0.76

UTH

0.75

 considerably

0.74

 athlet

0.73

 decisively

0.72

Activations Density 0.186%

phrases related to revealing or uncovering something hidden or unknown

metaphors related to openings and barriers

No Comments

No Known Activations