INDEX

Explanations

mentions of specific activities or events

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

Ã¨re

-0.88

bal

-0.80

si

-0.75

redo

-0.74

eka

-0.71

iere

-0.69

ifer

-0.67

 distinguishing

-0.67

rez

-0.66

respond

-0.65

POSITIVE LOGITS

redients

1.09

tons

0.88

 spree

0.79

bird

0.76

HAM

0.73

 engagements

0.71

lass

0.71

AME

0.70

 oneself

0.69

 aids

0.69

Activations Density 3.510%

mentions of specific activities or events

No Comments

No Known Activations