INDEX

Explanations

medical and pharmaceutical terms

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

LESS

-0.77

 Bauer

-0.73

nings

-0.73

 Wend

-0.71

çĶ°

-0.71

butt

-0.69

ãģĦ

-0.69

 Rapids

-0.67

boat

-0.67

EMENT

-0.65

POSITIVE LOGITS

ologically

1.44

odynamic

1.39

ogram

1.35

ologists

1.33

odynamics

1.33

ologic

1.33

ological

1.32

onyms

1.32

osph

1.31

ysics

1.30

Activations Density 2.113%

medical and pharmaceutical terms

No Comments

No Known Activations