INDEX

Explanations

names of entities, such as people, places, or organizations

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

ative

-0.80

atory

-0.74

rament

-0.70

á½

-0.67

ants

-0.65

ends

-0.65

atively

-0.62

ansk

-0.62

aries

-0.61

ary

-0.59

POSITIVE LOGITS

brook

0.90

ston

0.85

chief

0.78

ning

0.78

istor

0.77

structed

0.76

field

0.76

ner

0.76

nel

0.76

imation

0.75

Activations Density 1.899%

names of entities, such as people, places, or organizations

No Comments

No Known Activations