INDEX

Explanations

names and specific phrases related to personal interactions or disputes

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

Data Type

torch.float32

Hook Point

blocks.8.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

aneously

-0.66

istically

-0.65

ishers

-0.62

SIGN

-0.57

ously

-0.57

ctors

-0.56

ishment

-0.56

ASED

-0.56

ski

-0.55

ishly

-0.55

POSITIVE LOGITS

peed

1.56

ystem

1.53

hip

1.53

mith

1.48

aurus

1.46

erver

1.42

chool

1.42

hift

1.42

creen

1.41

pace

1.41

Activations Density 2.957%

names and specific phrases related to personal interactions or disputes

No Comments

No Known Activations