INDEX

Explanations

statements involving strong claims, warnings, and descriptions from various individuals

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

impact

-0.67

ija

-0.65

mite

-0.63

gradient

-0.62

aspx

-0.61

emaker

-0.60

mt

-0.59

ãĥİ

-0.58

veyard

-0.58

wayne

-0.57

POSITIVE LOGITS

 sarcast

0.82

 bluntly

0.76

 passionately

0.71

 apologised

0.71

:"

0.68

 remarks

0.68

 apolog

0.67

 apologise

0.66

 listeners

0.66

 forcefully

0.66

Activations Density 0.263%

statements involving strong claims, warnings, and descriptions from various individuals

No Comments

No Known Activations