INDEX

Explanations

phrases related to collaborations or agreements involving multiple parties

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

duc

-0.74

grim

-0.74

umer

-0.71

gun

-0.68

mington

-0.67

acco

-0.66

cheat

-0.66

den

-0.65

acher

-0.64

unc

-0.64

POSITIVE LOGITS

 between

1.23

between

1.05

 Between

1.05

 seamlessly

0.91

BET

0.90

 partners

0.88

 partner

0.87

SHIP

0.86

amorph

0.82

 dots

0.79

Activations Density 4.674%

phrases related to collaborations or agreements involving multiple parties

No Comments

No Known Activations