INDEX

Explanations

magic-related 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_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

dated

-0.69

etitive

-0.67

bis

-0.65

PA

-0.65

galitarian

-0.64

dit

-0.63

side

-0.63

ppers

-0.63

 Commit

-0.62

©¶æ

-0.61

POSITIVE LOGITS

 lantern

0.85

 wand

0.85

 Leap

0.83

ãĥĥãĤ¯

0.83

 mushrooms

0.81

 potion

0.76

 carpet

0.76

binding

0.76

 realism

0.76

 beasts

0.75

Activations Density 1.250%

magic-related terms

No Comments

No Known Activations