INDEX

Explanations

phrases related to essential items or processes

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

renheit

-0.81

Roy

-0.77

hare

-0.69

Ń·

-0.68

creen

-0.67

rows

-0.65

ells

-0.64

pard

-0.64

ammad

-0.64

ernels

-0.63

POSITIVE LOGITS

 ingredient

1.02

 prerequisite

0.95

isin

0.91

 components

0.85

 safeguard

0.84

 prec

0.83

 component

0.83

ogical

0.82

ists

0.82

ised

0.82

Activations Density 0.036%

phrases related to essential items or processes

No Comments

No Known Activations