Spaces:

tommytracx
/

thox-zero-general

Running on Zero

68.4 kB

Ctrl+K

2 contributors

ctx: raise MAX_CONTEXT_TOKENS 8192 -> 32768 (ThoxRoute registers these specialists at 256k; routed payloads of ~24k tokens were rejected by the 8k server)

c552c8d verified 4 days ago

.gitattributes

1.52 kB
initial commit 2 months ago
.gitignore

19 Bytes
Require an app-layer API key on /v1, and restore the output ceiling I dropped 2 months ago
HANDOFF.md

9.69 kB
Choose the quant by tier policy on measured evidence, not a hand-typed ladder 2 months ago
README.md

8.8 kB
Document the verified OpenAI conformance results 2 months ago
app.py

18.5 kB
Return OpenAI-shaped errors for malformed requests 2 months ago
llama_backend.py

16.1 kB
Require an app-layer API key on /v1, and restore the output ceiling I dropped 2 months ago
packages.txt

9 Bytes
Serve Qwen3.8-27B RVN Heretic GGUF via llama-server instead of Qwen3-8B via transformers 2 months ago
requirements.txt

774 Bytes
Serve Qwen3.8-27B RVN Heretic GGUF via llama-server instead of Qwen3-8B via transformers 2 months ago
thox_auth.py

4.85 kB
Return OpenAI-shaped errors for malformed requests 2 months ago
thox_space_config.py

8.11 kB
ctx: raise MAX_CONTEXT_TOKENS 8192 -> 32768 (ThoxRoute registers these specialists at 256k; routed payloads of ~24k tokens were rejected by the 8k server) 4 days ago