INDEX

Explanations

detailed descriptions on procedures or instructions

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

oubted

-0.66

room

-0.62

UM

-0.61

 ........

-0.60

iculture

-0.59

odder

-0.58

oubt

-0.57

piece

-0.57

peak

-0.56

esides

-0.56

POSITIVE LOGITS

soever

1.05

beit

0.96

ls

0.94

 much

0.94

itzer

0.90

ells

0.88

ever

0.83

ling

0.82

much

0.80

 MUCH

0.77

Activations Density 0.289%

detailed descriptions on procedures or instructions

No Comments

No Known Activations