INDEX

Explanations

communication and discussion related terms, potentially in informal settings like online forums

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

Events

-0.71

reports

-0.71

evidence

-0.70

Contents

-0.70

Sources

-0.69

 Atkins

-0.69

Edit

-0.67

overs

-0.67

Links

-0.66

AIDS

-0.66

POSITIVE LOGITS

lot

1.28

 bunch

1.25

uras

1.14

rouse

1.13

 handful

1.04

few

1.04

 couple

1.04

 multitude

1.04

 single

1.01

 particular

0.98

Activations Density 1.464%

communication and discussion related terms, potentially in informal settings like online forums

No Comments

No Known Activations