INDEX

Explanations

JavaScript code snippets with specific syntax patterns

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

 corrid

-0.77

 Anglic

-0.69

 Gothic

-0.67

 priesthood

-0.65

 satirical

-0.64

 Jama

-0.62

 Volvo

-0.62

arters

-0.62

 brill

-0.61

 Guantanamo

-0.61

POSITIVE LOGITS

defined

1.06

1.05

null

0.94

empty

0.92

true

0.85

==

0.85

existing

0.85

needed

0.82

 DEBUG

0.81

verified

0.80

Activations Density 0.015%

JavaScript code snippets with specific syntax patterns

No Comments

No Known Activations