INDEX

Explanations

words related to legal, political, and governmental topics

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

âĸ¬

-0.64

BUG

-0.63

asus

-0.63

VID

-0.63

ALLY

-0.62

atorium

-0.60

Args

-0.60

Õ

-0.59

ãĥį

-0.59

OV

-0.58

POSITIVE LOGITS

mith

1.61

pring

1.52

paces

1.48

pace

1.47

poons

1.45

hips

1.43

heet

1.37

hip

1.35

cale

1.35

etting

1.34

Activations Density 5.416%

words related to legal, political, and governmental topics

No Comments

No Known Activations