INDEX

Explanations

phrases related to conflict, controversy, or heated discussions

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

 Provision

-0.76

 sadd

-0.63

 Rockets

-0.62

Voy

-0.62

 Helm

-0.62

 Ethics

-0.61

 Sadd

-0.61

 Grail

-0.60

 Directorate

-0.58

 Bottom

-0.58

POSITIVE LOGITS

usal

1.09

iw

1.06

ocratic

1.06

umin

1.03

pron

1.01

aps

1.00

adic

1.00

ches

0.99

otropic

0.99

erb

0.99

Activations Density 1.975%

phrases related to conflict, controversy, or heated discussions

No Comments

No Known Activations