INDEX

Explanations

expressions of value or quality, particularly those that indicate something is good or beneficial

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rouse

-0.68

hov

-0.65

ogram

-0.64

orah

-0.63

ologue

-0.63

 scam

-0.62

 towed

-0.60

 brawl

-0.59

 roller

-0.59

ottest

-0.59

POSITIVE LOGITS

 distinctions

0.94

 commitments

0.94

 strides

0.91

 decisions

0.90

 copies

0.90

 allowances

0.87

 arrangements

0.84

 comparisons

0.83

 preparations

0.81

 concessions

0.80

Activations Density 0.038%

expressions of value or quality, particularly those that indicate something is good or beneficial

No Comments

No Known Activations