INDEX

Explanations

adverbs and phrases expressing evaluation or comparison

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

 Window

-0.71

 Gothic

-0.71

 consequential

-0.67

 Conversation

-0.66

aic

-0.66

 Meaning

-0.62

 Azerbai

-0.62

 Effective

-0.61

YING

-0.61

 Ange

-0.59

POSITIVE LOGITS

 took

1.22

 gave

1.16

 went

1.15

 blew

1.14

 underwent

1.13

 drove

1.13

 stole

1.12

got

1.11

 appeared

1.10

 knew

1.10

Activations Density 1.974%

adverbs and phrases expressing evaluation or comparison

No Comments

No Known Activations