INDEX

Explanations

specific named entities or proper nouns

oai_token-act-pair · gpt-3.5-turbo

conjunctions and phrases that signify connection or continuation in narratives

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

onica

-0.73

udding

-0.68

dding

-0.63

hoe

-0.63

oder

-0.61

undo

-0.61

Marginal

-0.61

°

-0.60

uction

-0.60

natureconservancy

-0.60

POSITIVE LOGITS

 consists

1.54

 contains

1.43

 comprises

1.36

 includes

1.35

 encompasses

1.24

 involves

1.23

 corresponds

1.23

 spans

1.23

 incorporates

1.20

 lasts

1.20

Activations Density 0.210%

specific named entities or proper nouns

conjunctions and phrases that signify connection or continuation in narratives

No Comments

No Known Activations