INDEX

Explanations

mentions of legal cases, court verdicts, and criminal activities

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

 roundup

-0.71

 outputs

-0.70

overs

-0.69

ocations

-0.68

 reproduction

-0.67

ysc

-0.66

alysis

-0.64

itures

-0.63

 progress

-0.62

olves

-0.61

POSITIVE LOGITS

Jr

1.52

Sr

1.34

Jr

1.26

who

1.25

who

1.23

 whose

1.13

aka

1.13

whose

1.13

 whom

1.10

 grandson

0.96

Activations Density 0.894%

mentions of legal cases, court verdicts, and criminal activities

No Comments

No Known Activations