eyad-silx commited on
Commit
6b153e4
·
verified ·
1 Parent(s): 81121cf
Files changed (1) hide show
  1. README.md +15 -0
README.md ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Ingot (SN97) challenger build — progress proof
2
+
3
+ Automated backup of build/train/eval artifacts from the H100 pod (RunPod, 1x H100 NVL).
4
+ Pushed by `hf_push.py` — checks hourly and uploads any new/changed artifact.
5
+
6
+ - `king-ix/` — incumbent king GGUF (Ingot-IX-27B-Q2) + provenance files (baseline)
7
+ - `chal-identity/`, `chal-v1/`, `chal-n2/`, `chal-norms/` — challenger GGUF iterations
8
+ (scale-refit / QAT variants; same 27B Q2 architecture as the king)
9
+ - `states/` — QAT state checkpoints (`t0..tN.safetensors`)
10
+ - `artifacts/` — `HANDOFF.md`, build/train/eval scripts, run logs, sample draws,
11
+ measurement results (`measure*.jsonl`, `pick*.jsonl`, `drawn.jsonl`)
12
+ - `MANIFEST.md` — size + sha256 + upload time of every artifact (regenerated each push)
13
+ - `PROGRESS.md` — push history
14
+
15
+ The GGUFs are ~6.8 GB each (PQ2_0 quantized 27B linear-attention models).