INDEX

Explanations

specific words related to a particular subject or category

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

BIL

-0.79

IVERS

-0.73

mol

-0.73

>>>

-0.73

hell

-0.72

perse

-0.68

-0.67

Adds

-0.67

ĺħ

-0.67

 ashore

-0.66

POSITIVE LOGITS

 kinds

1.16

ties

1.13

 subset

1.04

 wavelengths

1.03

 types

1.03

iating

1.00

 amount

1.00

 percentage

0.95

 geographic

0.95

 aspects

0.94

Activations Density 0.709%

specific words related to a particular subject or category

No Comments

No Known Activations