INDEX

Explanations

names or terms related to 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

ĸļ

-0.85

 enactment

-0.68

 exception

-0.67

ħĭ

-0.67

 caucuses

-0.65

ĪĴ

-0.65

ACTED

-0.64

 scrut

-0.59

 refere

-0.58

 consistency

-0.58

POSITIVE LOGITS

raltar

0.93

agall

0.90

ierrez

0.90

ppe

0.87

anamo

0.84

reau

0.82

protein

0.80

apons

0.78

vernment

0.77

intestinal

0.77

Activations Density 1.323%

names or terms related to individuals

No Comments

No Known Activations