INDEX

Explanations

words related to enjoyment, entertainment, and social interaction

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

rael

-0.66

 underest

-0.65

inx

-0.64

©¶æ

-0.63

 mater

-0.61

abases

-0.61

 bottleneck

-0.61

opic

-0.60

heed

-0.59

fixed

-0.59

POSITIVE LOGITS

nels

1.00

issance

0.97

sticks

0.83

oleon

0.81

 enjoyment

0.78

 Surprise

0.78

 stroll

0.78

tainment

0.77

Fest

0.77

osity

0.74

Activations Density 2.959%

words related to enjoyment, entertainment, and social interaction

No Comments

No Known Activations