INDEX

Explanations

recurring conjunctions and phrases in the context of relationships or comparisons

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

natureconservancy

-0.70

ORTS

-0.70

cial

-0.67

 Dickinson

-0.64

ãĤ´

-0.64

 Supplemental

-0.63

uren

-0.62

oren

-0.62

ATS

-0.61

urer

-0.60

POSITIVE LOGITS

 awhile

0.78

 forgotten

0.74

theless

0.70

isner

0.69

 regretted

0.69

estamp

0.67

 foremost

0.66

 limb

0.65

nown

0.65

thinkable

0.64

Activations Density 0.012%

recurring conjunctions and phrases in the context of relationships or comparisons

No Comments

No Known Activations