export step 1000 (3.47%)

README.md

@@ -13,11 +13,11 @@ tags:
 
 # Oleander-150M
 
-**Training checkpoint** (`export`) — step **6,500 / 28,820** · **22.55%** of the planned run.
+**Training checkpoint** (`export`) — step **1,000 / 28,820** · **3.47%** of the planned run.
 
 ```text
-steps    ██████░░░░░░░░░░░░░░░░░░░░░░  22.55%   6,500 / 28,820
-tokens   ██████░░░░░░░░░░░░░░░░░░░░░░  22.93%   3.46B / 15.11B compute
+steps    █░░░░░░░░░░░░░░░░░░░░░░░░░░░   3.47%   1,000 / 28,820
+tokens   █░░░░░░░░░░░░░░░░░░░░░░░░░░░   3.85%   580.91M / 15.11B compute
 phase    tst   (TST bag training ends at step 8,646)
 ```
 
@@ -27,17 +27,17 @@ Trained with the Oleander marimo trainer (molab): hybrid **GDN2** (gated delta-r
 
 | | |
 |---|---|
-| Step | **6,500** / 28,820 |
-| % complete (steps) | **22.55%** |
-| Compute tokens seen | **3.46B** / 15.11B (22.93%) |
-| Raw tokens seen (TST bags unpacked) | 20.79B / 37.77B |
-| Cluster tokens (all 1 worker(s)) | 3.46B / 15.11B compute |
+| Step | **1,000** / 28,820 |
+| % complete (steps) | **3.47%** |
+| Compute tokens seen | **580.91M** / 15.11B (3.85%) |
+| Raw tokens seen (TST bags unpacked) | 3.49B / 37.77B |
+| Cluster tokens (all 1 worker(s)) | 580.91M / 15.11B compute |
 | Phase | `tst` |
-| Val loss / ppl | **5.5231** / **250.40** |
-| Train loss (latest) | 5.2200 |
-| Throughput | 77,100 compute tok/s · 0.145 steps/s |
-| Session elapsed | 0d 11h 23m · ETA 1d 18h 53m |
-| Checkpoint written | 2026-11-13T07:27:27Z |
+| Val loss / ppl | **—** / **—** |
+| Train loss (latest) | 5.4966 |
+| Throughput | 76,839 compute tok/s · 0.126 steps/s |
+| Session elapsed | 0d 00h 55m · ETA 2d 13h 31m |
+| Checkpoint written | 2026-11-13T18:42:54Z |
 
 ## How to load
 
@@ -98,6 +98,6 @@ Oleander-150M (28L = 7 x [GDN2 · GDN2 · GDN2 · GQA] · d=576 · SwiGLU(1600)
 
 ## Notes
 
-- This is a **work-in-progress** checkpoint at 22.55% of the planned run, not a finished model: expect repetition and incoherence, especially with few-shot prompts.
+- This is a **work-in-progress** checkpoint at 3.47% of the planned run, not a finished model: expect repetition and incoherence, especially with few-shot prompts.
 - The tokenizer comes from the base model the run started from; the vocabulary is a plain BPE model and is not extended during training.
 - `training_meta.json` is regenerated on every push and is the authoritative record of how far training got.

config.json

@@ -113,12 +113,12 @@
   },
   "model_name": "Oleander-150M",
   "training": {
-    "step": 6500,
+    "step": 1000,
     "kind": "export",
     "max_steps": 28820,
-    "percent_complete": 22.554,
-    "tokens_seen": 20786970624,
-    "compute_tokens_seen": 3464495104
+    "percent_complete": 3.47,
+    "tokens_seen": 3485466624,
+    "compute_tokens_seen": 580911104
   },
   "transformers_version": "5.14.1",
   "torch_dtype": "float32"

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f8bb3242d79d74c0de4abc7396b103a854078451abc6282b1db01b065bd84985
+oid sha256:483b223660bed2cc9b713d67fb044729591cb087188ea28888cf4a552f49dd58
 size 599970860

progress.json

@@ -1,15 +1,15 @@
 {
-  "step": 6500,
+  "step": 1000,
   "max_steps": 28820,
-  "percent_complete": 22.554,
-  "tokens_seen": 20786970624,
-  "compute_tokens_seen": 3464495104,
+  "percent_complete": 3.47,
+  "tokens_seen": 3485466624,
+  "compute_tokens_seen": 580911104,
   "compute_tokens_total": 15109980160,
-  "percent_tokens": 22.929,
-  "val_loss": 5.523073740303516,
-  "val_ppl": 250.40353093471919,
+  "percent_tokens": 3.845,
+  "val_loss": null,
+  "val_ppl": null,
   "params_M": 149.976,
-  "pushed_at": "2026-11-13T07:27:27Z",
+  "pushed_at": "2026-11-13T18:42:54Z",
   "architecture": "Oleander-150M",
   "features": [
     "GDN2 (gated delta net 2) x21: 9h x 64, conv4",

training_meta.json

@@ -1,10 +1,10 @@
 {
   "format": "oleander-training-meta/1",
-  "generated_at": "2026-11-13T07:27:27Z",
+  "generated_at": "2026-11-13T18:42:54Z",
   "kind": "export",
-  "step": 6500,
-  "tokens_seen": 20786970624,
-  "pushed_at": "2026-11-13T07:27:27Z",
+  "step": 1000,
+  "tokens_seen": 3485466624,
+  "pushed_at": "2026-11-13T18:42:54Z",
   "params_M": 149.976,
   "architecture": "Oleander-150M",
   "features": [
@@ -20,55 +20,55 @@
     "TST (token superposition training, bag 6)"
   ],
   "optimizer": "muon_adamw",
-  "val_loss": 5.523073740303516,
-  "val_ppl": 250.40353093471919,
+  "val_loss": null,
+  "val_ppl": null,
   "run": {
     "model_name": "Oleander-150M",
     "architecture_detail": "Oleander-150M (28L = 7 x [GDN2 \u00b7 GDN2 \u00b7 GDN2 \u00b7 GQA] \u00b7 d=576 \u00b7 SwiGLU(1600) \u00b7 tied 32,770 vocab)",
-    "session_id": "20261112_195408_1a765b",
+    "session_id": "20261113_174058_13edde",
     "repo_id": "CyanMonkey/Test-Subject-3",
-    "started_at": "2026-11-12T20:04:22Z",
-    "session_elapsed_s": 40985.1,
-    "last_update_at": "2026-11-13T07:27:25Z",
-    "steps_per_second": 0.14454,
-    "compute_tokens_per_second": 77099.8,
-    "eta": "1d 18h 53m",
+    "started_at": "2026-11-13T17:47:10Z",
+    "session_elapsed_s": 3343.6,
+    "last_update_at": "2026-11-13T18:42:48Z",
+    "steps_per_second": 0.12559,
+    "compute_tokens_per_second": 76839.1,
+    "eta": "2d 13h 31m",
     "checkpoint_kind": "export"
   },
   "progress": {
-    "step": 6500,
+    "step": 1000,
     "max_steps": 28820,
-    "step_fraction": 0.225538,
-    "percent_complete": 22.554,
-    "tokens_fraction": 0.229285,
-    "percent_tokens": 22.929,
+    "step_fraction": 0.034698,
+    "percent_complete": 3.47,
+    "tokens_fraction": 0.038446,
+    "percent_tokens": 3.845,
     "phase": "tst",
-    "phase_fraction": 0.751793,
+    "phase_fraction": 0.11566,
     "tst_switch_step": 8646,
     "tst_switch_applied": false
   },
   "tokens": {
-    "compute_tokens_seen": 3464495104,
+    "compute_tokens_seen": 580911104,
     "compute_tokens_total": 15109980160,
-    "raw_tokens_seen": 20786970624,
+    "raw_tokens_seen": 3485466624,
     "raw_tokens_total": 37774950400,
     "raw_over_compute_ratio": 6.0,
-    "samples_seen": 19968000,
-    "rows_consumed": 19968000,
+    "samples_seen": 3072000,
+    "rows_consumed": 3072000,
     "compute_tokens_per_step": 524288,
     "cluster": {
       "n_workers": 1,
-      "compute_tokens_seen_all_workers": 3464495104,
-      "raw_tokens_seen_all_workers": 20786970624,
+      "compute_tokens_seen_all_workers": 580911104,
+      "raw_tokens_seen_all_workers": 3485466624,
       "compute_tokens_total_planned_all_workers": 15109980160,
       "raw_tokens_total_planned_all_workers": 37774950400
     }
   },
   "metrics": {
-    "train_loss": 5.220016002655029,
-    "val_loss": 5.523073740303516,
-    "val_ppl": 250.40353093471919,
-    "grad_norm": 0.3239007890224457,
+    "train_loss": 5.496631145477295,
+    "val_loss": null,
+    "val_ppl": null,
+    "grad_norm": 0.2176198810338974,
     "lr_mult": 1.0
   },
   "optimization": {
@@ -151,7 +151,7 @@
     "doc_masking": "auto",
     "shuffle_block_rows": 192,
     "seed": 42,
-    "epochs_consumed_this_worker": 1.3534
+    "epochs_consumed_this_worker": 0.2082
   },
   "hardware": {
     "device": "cuda",