INDEX

Explanations

terms related to evaluating or re-evaluating something

oai_token-act-pair · gpt-3.5-turbo

terms related to evaluation and analysis processes

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 spare

-0.76

 Tribe

-0.73

 Johnson

-0.69

 giveaway

-0.64

 lead

-0.63

BJ

-0.62

 Larson

-0.62

rob

-0.61

 sake

-0.61

TD

-0.61

POSITIVE LOGITS

eval

4.41

analysis

1.91

evaluate

1.52

 eval

1.37

analy

1.36

examination

1.25

educ

1.14

 Eval

1.10

classified

1.04

ework

1.00

Activations Density 0.029%

terms related to evaluating or re-evaluating something

terms related to evaluation and analysis processes

No Comments

No Known Activations