INDEX

Explanations

negations or opposing phrases and their relationships in context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

 gestation

-0.78

eger

-0.73

 Juda

-0.60

 mitigation

-0.59

 satisf

-0.59

 faults

-0.58

 corridors

-0.58

 fault

-0.57

 lowest

-0.57

hov

-0.57

POSITIVE LOGITS

thood

0.74

chio

0.69

elly

0.69

 Wolver

0.67

sense

0.66

asty

0.65

bang

0.64

nan

0.64

money

0.63

iz

0.63

Activations Density 0.027%

negations or opposing phrases and their relationships in context

No Comments

No Known Activations