INDEX

Explanations

phrases and words that convey a sense of revelation or highlight significant contrasts

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

ategory

-1.02

gres

-0.81

hops

-0.81

neys

-0.79

ulhu

-0.79

lished

-0.79

Jump

-0.77

edia

-0.76

ONEY

-0.73

afety

-0.72

POSITIVE LOGITS

 resemblance

1.01

ly

0.94

 likeness

0.93

 understatement

0.88

 resemb

0.84

 contrasts

0.84

 similarity

0.82

 illustration

0.81

tale

0.81

 indictment

0.79

Activations Density 0.024%

phrases and words that convey a sense of revelation or highlight significant contrasts

No Comments

No Known Activations