INDEX

Explanations

references to physical holes or voids

oai_token-act-pair · gpt-3.5-turbo

references to "holes" in various contexts, including physical, metaphorical, and structural meanings

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Bal

-0.81

lux

-0.69

NX

-0.68

 Riding

-0.62

cia

-0.60

uno

-0.60

against

-0.59

ority

-0.59

vez

-0.58

IGHTS

-0.58

POSITIVE LOGITS

 drilled

1.28

 hole

1.16

 holes

1.14

dug

1.14

hole

1.07

 Hole

1.05

 punched

1.04

holes

0.95

pots

0.92

 poked

0.91

Activations Density 0.082%

references to physical holes or voids

references to "holes" in various contexts, including physical, metaphorical, and structural meanings

No Comments

No Known Activations