INDEX

Explanations

information related to weights and measurements

oai_token-act-pair · gpt-3.5-turbo

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

 Painter

-0.80

 sunset

-0.78

 pathway

-0.75

 deepening

-0.71

 urgency

-0.69

Gate

-0.66

 chorus

-0.65

 corridor

-0.65

 acad

-0.64

 hotline

-0.63

POSITIVE LOGITS

 kilograms

1.66

kg

1.61

lbs

1.59

kg

1.58

lbs

1.57

 pounds

1.57

 ounces

1.40

lb

1.36

 grams

1.36

kil

1.34

Activations Density 0.095%

information related to weights and measurements

No Comments

No Known Activations