Add files using upload-large-folder tool

README.md

@@ -43,7 +43,7 @@ The main model repository is intended to expose the latest model-only artifact:
 
 It is not intended to keep every training checkpoint as visible model files. Intermediate FSDP2 `.distcp` checkpoints are large resume artifacts and are kept separately in `LLM-OS-Models/KoHRM-Text-1.4B-raw-checkpoints` when needed. The main repo may still have normal Hugging Face git history, but the current file tree should be treated as the latest public model export.
 
-Current public artifact: `stage1` HRM fast-cap checkpoint at `step_25000`, converted with EMA weights to `safetensors`. Training is still in progress.
+Current public artifact: `stage2` HRM full/no-cap checkpoint at `step_120000`, converted with EMA weights to `safetensors`. Training is still in progress from this run.
 
 ## Model Details
 
@@ -117,7 +117,8 @@ from simple_inference_engine import inference_load_checkpoint, inference_generat
 
 ckpt = inference_load_checkpoint(
     ckpt_path="/path/to/KoHRM-Text-1.4B-stage1-hrm-fastcap-gbs180",
-    ckpt_epoch=25000,
+    ckpt_epoch=None,
+    ckpt_step=85000,
     ckpt_use_ema=True,
     device="cuda",
 )
@@ -156,7 +157,7 @@ Completed and prepared datasets:
 | SWE-ZERO + GLM pilot mix | 251.2M | 990M | included in stage-0 mix |
 | Korean legal SFT/task data | 83.1M | 336M | included in stage-0 mix |
 | ToolBench train tool-call data | 127.0M | 500M | included in stage-0 mix |
-| HRM cleaned fast-cap stage-1 | 14.55B | 148G | current stage-1 |
+| HRM cleaned fast-cap stage-1 | 14.55B | 148G | completed through latest saved `step_85000` |
 | Korean statutes/local ordinances raw full | 308.9M | 1.2G | prepared for later stages |
 | Korean administrative rules + precedents raw full | 271.7M | 1.1G | prepared for later stages |
 | Korean legal/admin full task data | 629.0M | 2.5G | uploaded to prepared dataset repo |
@@ -207,14 +208,15 @@ The current public checkpoint was produced through staged pretraining:
 1. Train `stage-0` on `koterm_pretrain_mix_v1` with 711.3M tokens.
 2. Continue once more on the same available mix as `stage0b`.
 3. Continue to `stage-1` on HRM cleaned fast-cap data with 14.55B tokens.
-4. Convert `stage1 step_25000` EMA weights to `safetensors` and upload to the main model repo.
+4. Convert `stage2 step_120000` EMA weights to `safetensors` and upload to the main model repo.
+5. Continue from `stage1 step_85000` into `stage2` on full/no-cap HRM cleaned data.
 
-Current long-running stage-1 settings:
+Current long-running stage-2 settings:
 
 | Field | Value |
 |---|---|
 | Hardware | 8 x NVIDIA H200 |
-| Data | `koterm_hrm_cleaned_fastcap_stage1_v1` |
+| Data | `koterm_hrm_cleaned_full_nocap_v1` |
 | Tokens in current stage dataset | 14.55B |
 | Global batch | 180,224 tokens |
 | Local token slots/GPU | 22,528 |

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:46d4e7e588bf66dc44aa036c7ed4542428d7832de0ae330288486e0c19370043
+oid sha256:27fbd40942654580e230a7bc20cbd7c1019cc571f2c42d54a3e6985a21cd545d
 size 2768259784