INDEX

Explanations

mentions of physical compartments or separations

oai_token-act-pair · gpt-3.5-turbo

terms related to dividing or partitioning entities

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

mens

-0.90

NG

-0.85

 Hunts

-0.80

Nex

-0.77

Tech

-0.76

tal

-0.75

onga

-0.74

{{

-0.72

 Nigerian

-0.71

tech

-0.71

POSITIVE LOGITS

 partition

3.90

 partitions

3.03

 compartment

1.33

 fragmentation

1.26

 Shank

1.06

 pane

1.02

 filesystem

1.01

Disk

1.00

 entropy

0.98

 divid

0.98

Activations Density 0.037%

mentions of physical compartments or separations

terms related to dividing or partitioning entities

No Comments

No Known Activations