INDEX

Explanations

expressions of personal feelings or emotions

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

 Camer

-0.74

adish

-0.66

 Display

-0.63

 Tactics

-0.60

 Accounting

-0.59

esan

-0.59

 Prospect

-0.58

angle

-0.58

 Lies

-0.58

photos

-0.58

POSITIVE LOGITS

 comfortable

1.07

 compelled

1.06

 uneasy

1.04

 uncomfortable

0.92

 confident

0.91

 obligated

0.90

 vibrations

0.90

 strongly

0.89

 sorry

0.85

 betrayed

0.83

Activations Density 0.381%

expressions of personal feelings or emotions

No Comments

No Known Activations