INDEX

Explanations

keywords related to goals and aims

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

 anomalies

-0.73

 Solitaire

-0.70

 alloy

-0.69

ania

-0.67

å£«

-0.64

 bruises

-0.63

ols

-0.62

 examples

-0.62

âĶĢâĶĢâĶĢâĶĢ

-0.61

parts

-0.60

POSITIVE LOGITS

Goal

1.11

 undertaken

0.98

 aimed

0.95

 Goal

0.94

 toward

0.93

goal

0.92

 goal

0.92

 pursued

0.91

irection

0.87

 goals

0.87

Activations Density 5.150%

keywords related to goals and aims

No Comments

No Known Activations