INDEX

Explanations

references to colors and color combinations

oai_token-act-pair · gpt-3.5-turbo

references to colors

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 logistics

-0.77

EH

-0.73

odcast

-0.72

endon

-0.68

ository

-0.68

 servicing

-0.67

 disposal

-0.67

SER

-0.67

 disse

-0.67

=-=-=-=-

-0.66

POSITIVE LOGITS

 colors

2.00

 color

1.90

 colours

1.84

 colour

1.78

 violet

1.67

 purple

1.67

 Colors

1.66

hue

1.65

 orange

1.64

 pink

1.60

Activations Density 0.501%

references to colors and color combinations

references to colors

No Comments

No Known Activations