INDEX

Explanations

references to physical holes or voids

oai_token-act-pair · gpt-3.5-turbo

references to holes or voids

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 proc

-0.65

att

-0.63

 renewed

-0.62

 Import

-0.61

 unic

-0.61

 Camb

-0.60

 ACPI

-0.60

res

-0.60

 spont

-0.60

-0.59

POSITIVE LOGITS

hole

4.82

holes

4.50

 hole

2.05

 Hole

1.85

 holes

1.74

hol

1.61

HO

1.07

pie

1.03

hower

1.02

pots

1.00

Activations Density 0.013%

references to physical holes or voids

references to holes or voids

No Comments

No Known Activations