INDEX

Explanations

names and entities related to various topics such as sports, news outlets, and specific individuals

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

 Ninth

-0.66

 jaws

-0.66

 shrink

-0.64

 backward

-0.63

Goo

-0.63

 Goldberg

-0.63

 Scheme

-0.63

 cumbers

-0.62

 rounding

-0.62

ABE

-0.62

POSITIVE LOGITS

1.49

/)

1.25

__

1.22

_-_

1.18

/,

1.17

_(

1.09

reports

1.06

_.

1.03

/?

1.02

/.

1.02

Activations Density 1.238%

names and entities related to various topics such as sports, news outlets, and specific individuals

No Comments

No Known Activations