Decision Workspace
llm-samplers vs ghost-flow vs ggml-sys-bleedingedge
Side-by-side comparison of Rust crates
42
llm-samplers
growingv0.0.7
Token samplers for large language models
54
ghost-flow
experimentalv1.7.0
Complete machine learning framework in Rust - tensors, neural networks, 85+ ML algorithms, GPU acceleration, WASM, FFI, model serving, compiler optimizations, distributed training, and 8 hardware accelerators - all included by default
51
ggml-sys-bleedingedge
growingv2404291813.0.0+llamacpp-release.b2757
Bleeding edge low-level bindings to GGML.
Core Metrics
| llm-samplers | ghost-flow | ggml-sys-bleedingedge | |
|---|---|---|---|
| Health Score | 42 | 54 | 51 |
| Total Downloads | 118.8K | 147 | 683.8K |
| 30d Downloads | 3.2K | 0 | 0 |
| Dependents | 48 | 0 | 6 |
| Releases | 6 | 8 | 493 |
| Last Updated | 928d ago | 126d ago | 756d ago |
| Age | 2y 11m | 4m | 3y 1m |
Health Breakdown
llm-samplers
Maintenance
4
Quality
14
Community
8
Popularity
6
Documentation
10
ghost-flow
Maintenance
13
Quality
19
Community
6
Popularity
3
Documentation
13
ggml-sys-bleedingedge
Maintenance
10
Quality
19
Community
5
Popularity
7
Documentation
10
Technical Details
| llm-samplers | ghost-flow | ggml-sys-bleedingedge | |
|---|---|---|---|
| Version | 0.0.7 | 1.7.0 | 2404291813.0.0+llamacpp-release.b2757 |
| Stable (≥1.0) | ✗ No | ✓ Yes | ✓ Yes |
| License | MIT | MIT OR Apache-2.0 | MIT |
| Dependencies | 4 | 10 | 3 |
| Crate Size | 28KB | 23KB | 9.0MB |
| Features | 0 | 12 | 10 |
| Yanked % | 0.0% | 0.0% | 0.4% |
| Edition | 2021 | 2021 | 2021 |
| MSRV | — | — | — |
| Owners | 1 | 1 | 1 |
Links
Quick Verdict
- •ghost-flow leads with a health score of 54/100, but none of the options score above 80.
- •ggml-sys-bleedingedge has the most downloads (683.8K), suggesting wider adoption.
- •⚠ llm-samplers, ggml-sys-bleedingedge have not been updated in over a year.
- •llm-samplers is pre-1.0 — API may change.