dist.dirigoagents.com — BitTorrent model mirror
License-cleared GGUF language models, distributed as a peer-accelerated BitTorrent
swarm with an operator-signed catalog and an HTTP webseed fallback. Files are served
by the tracker at dist.dirigoagents.com:6969 and mirrored over HTTPS/HTTP.
A 0.5-billion-parameter, instruction-tuned language model from Alibaba's Qwen2.5 family, quantized to 8-bit (Q8_0) for near-lossless quality at minimal footprint. Ideal for lightweight local chat, instruction following, and basic tool / function-calling agent prototypes on constrained hardware.
| Property | Value |
|---|---|
| Model | qwen2.5-0.5b-instruct-q8_0.gguf |
| Parameters | 0.5 B |
| Quantization | Q8_0 (8-bit) |
| File size | 675,710,816 bytes (644.4 MiB) |
| SHA-256 | ca59ca7f13d0e15a8cfa77bd17e65d24f6844b554a7b6c12e07a5f89ff76844e |
| Info hash (btih) | b50ba63ed20072061df6d7ec577038ad8b93669f |
| License | Apache-2.0 (Qwen2.5) |
aria2c 'magnet:?xt=urn:btih:b50ba63ed20072061df6d7ec577038ad8b93669f&dn=qwen2.5-0.5b-instruct-q8_0&tr=http%3A%2F%2Fdist.dirigoagents.com%3A6969%2Fannounce&ws=http%3A%2F%2Fdist.dirigoagents.com%2Ffiles%2FLLMs%2Fqwen2.5-0.5b-instruct-q8_0.gguf&ws=https%3A%2F%2Fhuggingface.co%2FQwen%2FQwen2.5-0.5B-Instruct-GGUF%2Fresolve%2Fmain%2Fqwen2.5-0.5b-instruct-q8_0.gguf'
Every file is vetted and listed in an ed25519-signed operator catalog; clients verify the signature before downloading and re-check SHA-256 after. The swarm is seeded continuously from this host, with an HTTP webseed as a fallback when peer traffic is blocked. More models are added to this catalog as operator storage expands.