INDEX

Explanations

instances related to self-reflection, personal experience, and self-identity

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

 XIII

-0.71

heny

-0.68

cheon

-0.68

SHIP

-0.68

rice

-0.68

ī

-0.67

ondo

-0.66

 Ashe

-0.66

endar

-0.64

yk

-0.63

POSITIVE LOGITS

destruct

1.08

conscious

0.88

same

0.87

lessly

0.85

 destruct

0.82

proclaimed

0.82

 explanatory

0.81

 esteem

0.79

 conscious

0.76

pres

0.75

Activations Density 5.191%

instances related to self-reflection, personal experience, and self-identity

No Comments

No Known Activations