INDEX

Explanations

medical or health-related terms, especially those related to mental health or conditions requiring care and protection, along with terms related to societal issues and controversies

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

 briefs

-0.77

OPLE

-0.72

MX

-0.67

 Architecture

-0.66

azine

-0.65

phrine

-0.64

 Blitz

-0.64

 Dynamics

-0.63

 simulations

-0.63

 retreat

-0.63

POSITIVE LOGITS

assuming

1.25

balanced

1.24

ruly

1.20

readable

1.18

numbered

1.18

earned

1.15

ifying

1.15

qualified

1.14

animous

1.12

ortunately

1.12

Activations Density 2.163%

medical or health-related terms, especially those related to mental health or conditions requiring care and protection, along with terms related to societal issues and controversies

No Comments

No Known Activations