INDEX

Explanations

similes comparing actions or events to the way something else behaves

oai_token-act-pair · gpt-3.5-turbo

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

ulic

-0.70

hirt

-0.69

qua

-0.68

inas

-0.68

SEE

-0.67

VICE

-0.66

ilic

-0.65

inge

-0.65

icipated

-0.65

hetics

-0.64

POSITIVE LOGITS

 wildfire

1.20

 crazy

1.05

 clock

0.97

 bandits

0.86

mad

0.86

liest

0.84

 weeds

0.84

 flies

0.84

 glue

0.82

 rabbits

0.82

Activations Density 0.053%

similes comparing actions or events to the way something else behaves

No Comments

No Known Activations