INDEX

Explanations

names and terms related to individuals or organizations

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

 faint

-0.73

hyde

-0.69

 congr

-0.67

 lapse

-0.66

 inaug

-0.65

 accent

-0.64

 Recomm

-0.63

 subt

-0.61

 Samar

-0.60

REDACTED

-0.59

POSITIVE LOGITS

hett

0.90

idon

0.90

cheon

0.85

tle

0.82

itzer

0.81

oslav

0.76

aughters

0.75

hemy

0.74

ifix

0.73

cross

0.73

Activations Density 1.313%

names and terms related to individuals or organizations

No Comments

No Known Activations