INDEX

Explanations

numerical sequences in a specific format

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 util

-0.95

 inclusion

-0.76

 stated

-0.74

FG

-0.74

tv

-0.73

 ability

-0.73

due

-0.72

 phased

-0.71

 oath

-0.71

 enqu

-0.70

POSITIVE LOGITS

Advertisement

1.84

Newsletter

1.82

Yet

1.68

Such

1.57

Still

1.56

But

1.54

Whatever

1.50

Meanwhile

1.49

Across

1.46

Neither

1.46

Activations Density 0.481%

numerical sequences in a specific format

No Comments

No Known Activations