INDEX

Explanations

references to specific capitalized words or proper nouns within the text

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

ritch

-0.66

odox

-0.66

anted

-0.64

verified

-0.63

qualified

-0.63

tmp

-0.62

 Canaver

-0.61

pta

-0.61

acqu

-0.61

yrim

-0.61

POSITIVE LOGITS

(>

0.82

Thumbnail

0.76

 daddy

0.70

gie

0.68

————————

0.68

buster

0.67

 lumber

0.67

 scoop

0.65

TVs

0.64

 swoop

0.64

Activations Density 0.100%

references to specific capitalized words or proper nouns within the text

No Comments

No Known Activations