INDEX

Explanations

words related to attention or emphasis

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

ania

-0.80

asca

-0.75

idden

-0.70

named

-0.68

OUGH

-0.67

ston

-0.65

mia

-0.62

ccess

-0.62

 paran

-0.62

ibrary

-0.61

POSITIVE LOGITS

starter

0.88

rite

0.86

Goal

0.82

 squarely

0.82

 focuses

0.80

ivation

0.80

 focus

0.79

fulness

0.78

 focused

0.78

 toward

0.74

Activations Density 0.025%

words related to attention or emphasis

No Comments

No Known Activations