INDEX

Explanations

names or references to individuals and their actions

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

 Trave

-0.67

 Daly

-0.66

ishable

-0.64

 Brisbane

-0.59

 Millennials

-0.57

 Berry

-0.57

 Sovereign

-0.56

Fla

-0.56

cientious

-0.56

 Burg

-0.56

POSITIVE LOGITS

 consisted

0.88

 belonged

0.76

 reportedly

0.76

anyahu

0.76

was

0.76

 testified

0.76

 specializes

0.74

 wasn

0.74

 alone

0.73

 consists

0.73

Activations Density 2.348%

names or references to individuals and their actions

No Comments

No Known Activations