INDEX

Explanations

occurrences of specific numbers and punctuation

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

inement

-0.73

 pursu

-0.72

 transgress

-0.66

æ©

-0.65

 targeted

-0.64

 skirm

-0.64

 overl

-0.64

aband

-0.63

 equivalents

-0.63

 arous

-0.63

POSITIVE LOGITS

 Yeah

1.30

 Alright

1.12

 Okay

1.01

 Exactly

1.01

 Originally

0.99

 Absolutely

0.98

Yeah

0.97

Hmm

0.94

Hey

0.93

 Firstly

0.93

Activations Density 0.050%

occurrences of specific numbers and punctuation

No Comments

No Known Activations