INDEX

Explanations

phrases related to groups of people or general statements about 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_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

tnc

-0.79

iger

-0.75

ult

-0.69

ories

-0.65

urations

-0.64

 Resurrection

-0.63

urer

-0.61

FU

-0.60

urrection

-0.60

QL

-0.60

POSITIVE LOGITS

 else

2.01

 Else

1.42

else

1.40

Else

1.30

who

1.04

 knows

0.93

 acquainted

0.90

 involved

0.84

 remembers

0.82

 alike

0.82

Activations Density 0.431%

phrases related to groups of people or general statements about individuals

No Comments

No Known Activations