INDEX

Explanations

numbers or values that have exceeded or surpassed certain benchmarks

oai_token-act-pair · gpt-3.5-turbo

terms related to exceeding or surpassing numerical thresholds or quantities

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.10.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

atto

-0.77

lean

-0.72

BUG

-0.68

Secure

-0.66

otta

-0.63

Mesh

-0.63

legged

-0.63

onna

-0.61

aci

-0.61

lique

-0.61

POSITIVE LOGITS

 expectations

1.36

 predictions

0.91

0.90

 milestones

0.84

 projections

0.84

0.83

 expectation

0.82

0.81

 forecasts

0.80

Activations Density 0.079%

numbers or values that have exceeded or surpassed certain benchmarks

terms related to exceeding or surpassing numerical thresholds or quantities

No Comments

No Known Activations