INDEX

Explanations

references to common sense reasoning

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

atern

-0.90

leased

-0.90

usky

-0.79

Stars

-0.79

phies

-0.78

reon

-0.71

RAW

-0.70

semble

-0.65

classified

-0.65

\/\/

-0.64

POSITIVE LOGITS

 prevailed

0.96

smanship

0.90

 prev

0.90

 dictates

0.90

 prevail

0.85

 decency

0.82

 argument

0.80

 approach

0.79

 judgement

0.76

 advice

0.74

Activations Density 0.020%

references to common sense reasoning

No Comments

No Known Activations