INDEX

Explanations

numerical data in the format of numbers and percentages

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

Joy

-0.77

gone

-0.67

Minnesota

-0.66

than

-0.65

NAS

-0.64

 Polk

-0.61

laws

-0.61

hoe

-0.61

Marginal

-0.60

jan

-0.59

POSITIVE LOGITS

 depending

1.23

 inclusive

1.02

depending

0.98

 respectively

0.94

 alike

0.78

 transsexual

0.77

)=(

0.74

 ratio

0.69

oldemort

0.69

 metab

0.67

Activations Density 2.079%

numerical data in the format of numbers and percentages

No Comments

No Known Activations