CyanMonkey commited on
Commit
ffb4be2
·
verified ·
1 Parent(s): f84ab14

export step 5500 (19.08%)

Browse files
Files changed (5) hide show
  1. README.md +14 -14
  2. config.json +4 -4
  3. model.safetensors +1 -1
  4. progress.json +8 -8
  5. training_meta.json +28 -28
README.md CHANGED
@@ -13,11 +13,11 @@ tags:
13
 
14
  # Oleander-150M
15
 
16
- **Training checkpoint** (`export`) — step **5,000 / 28,820** · **17.35%** of the planned run.
17
 
18
  ```text
19
- steps █████░░░░░░░░░░░░░░░░░░░░░░░ 17.35% 5,000 / 28,820
20
- tokens █████░░░░░░░░░░░░░░░░░░░░░░░ 17.72% 2.68B / 15.11B compute
21
  phase tst (TST bag training ends at step 8,646)
22
  ```
23
 
@@ -27,17 +27,17 @@ Trained with the Oleander marimo trainer (molab): hybrid **GDN2** (gated delta-r
27
 
28
  | | |
29
  |---|---|
30
- | Step | **5,000** / 28,820 |
31
- | % complete (steps) | **17.35%** |
32
- | Compute tokens seen | **2.68B** / 15.11B (17.72%) |
33
- | Raw tokens seen (TST bags unpacked) | 16.07B / 37.77B |
34
- | Cluster tokens (all 1 worker(s)) | 2.68B / 15.11B compute |
35
  | Phase | `tst` |
36
- | Val loss / ppl | **5.6935** / **296.94** |
37
- | Train loss (latest) | 5.3230 |
38
- | Throughput | 76,857 compute tok/s · 0.144 steps/s |
39
- | Session elapsed | 0d 08h 33m · ETA 1d 22h 05m |
40
- | Checkpoint written | 2026-11-14T02:20:42Z |
41
 
42
  ## How to load
43
 
@@ -98,6 +98,6 @@ Oleander-150M (28L = 7 x [GDN2 · GDN2 · GDN2 · GQA] · d=576 · SwiGLU(1600)
98
 
99
  ## Notes
100
 
101
- - This is a **work-in-progress** checkpoint at 17.35% of the planned run, not a finished model: expect repetition and incoherence, especially with few-shot prompts.
102
  - The tokenizer comes from the base model the run started from; the vocabulary is a plain BPE model and is not extended during training.
103
  - `training_meta.json` is regenerated on every push and is the authoritative record of how far training got.
 
13
 
14
  # Oleander-150M
15
 
16
+ **Training checkpoint** (`export`) — step **5,500 / 28,820** · **19.08%** of the planned run.
17
 
18
  ```text
19
+ steps █████░░░░░░░░░░░░░░░░░░░░░░░ 19.08% 5,500 / 28,820
20
+ tokens █████░░░░░░░░░░░░░░░░░░░░░░░ 19.46% 2.94B / 15.11B compute
21
  phase tst (TST bag training ends at step 8,646)
22
  ```
23
 
 
27
 
28
  | | |
29
  |---|---|
30
+ | Step | **5,500** / 28,820 |
31
+ | % complete (steps) | **19.08%** |
32
+ | Compute tokens seen | **2.94B** / 15.11B (19.46%) |
33
+ | Raw tokens seen (TST bags unpacked) | 17.64B / 37.77B |
34
+ | Cluster tokens (all 1 worker(s)) | 2.94B / 15.11B compute |
35
  | Phase | `tst` |
36
+ | Val loss / ppl | **5.9075** / **367.79** |
37
+ | Train loss (latest) | 5.2668 |
38
+ | Throughput | 76,882 compute tok/s · 0.144 steps/s |
39
+ | Session elapsed | 0d 09h 30m · ETA 1d 21h 03m |
40
+ | Checkpoint written | 2026-11-14T03:17:54Z |
41
 
42
  ## How to load
43
 
 
98
 
99
  ## Notes
100
 
101
+ - This is a **work-in-progress** checkpoint at 19.08% of the planned run, not a finished model: expect repetition and incoherence, especially with few-shot prompts.
102
  - The tokenizer comes from the base model the run started from; the vocabulary is a plain BPE model and is not extended during training.
103
  - `training_meta.json` is regenerated on every push and is the authoritative record of how far training got.
config.json CHANGED
@@ -113,12 +113,12 @@
113
  },
114
  "model_name": "Oleander-150M",
115
  "training": {
116
- "step": 5000,
117
  "kind": "export",
118
  "max_steps": 28820,
119
- "percent_complete": 17.349,
120
- "tokens_seen": 16068378624,
121
- "compute_tokens_seen": 2678063104
122
  },
123
  "transformers_version": "5.14.1",
124
  "torch_dtype": "float32"
 
113
  },
114
  "model_name": "Oleander-150M",
115
  "training": {
116
+ "step": 5500,
117
  "kind": "export",
118
  "max_steps": 28820,
119
+ "percent_complete": 19.084,
120
+ "tokens_seen": 17641242624,
121
+ "compute_tokens_seen": 2940207104
122
  },
123
  "transformers_version": "5.14.1",
124
  "torch_dtype": "float32"
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:eebef458e461899cf19678b76dfe8e2d812ce494189fca78f4e1c27239e49f68
3
  size 599970860
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a7e1044aa0634f89aed572a4f4b26a0a277dd8eca7d4f3f80165a5ed1d62f872
3
  size 599970860
progress.json CHANGED
@@ -1,15 +1,15 @@
1
  {
2
- "step": 5000,
3
  "max_steps": 28820,
4
- "percent_complete": 17.349,
5
- "tokens_seen": 16068378624,
6
- "compute_tokens_seen": 2678063104,
7
  "compute_tokens_total": 15109980160,
8
- "percent_tokens": 17.724,
9
- "val_loss": 5.6935174986720085,
10
- "val_ppl": 296.93625872214756,
11
  "params_M": 149.976,
12
- "pushed_at": "2026-11-14T02:20:42Z",
13
  "architecture": "Oleander-150M",
14
  "features": [
15
  "GDN2 (gated delta net 2) x21: 9h x 64, conv4",
 
1
  {
2
+ "step": 5500,
3
  "max_steps": 28820,
4
+ "percent_complete": 19.084,
5
+ "tokens_seen": 17641242624,
6
+ "compute_tokens_seen": 2940207104,
7
  "compute_tokens_total": 15109980160,
8
+ "percent_tokens": 19.459,
9
+ "val_loss": 5.9075037613511086,
10
+ "val_ppl": 367.78692464115176,
11
  "params_M": 149.976,
12
+ "pushed_at": "2026-11-14T03:17:54Z",
13
  "architecture": "Oleander-150M",
14
  "features": [
15
  "GDN2 (gated delta net 2) x21: 9h x 64, conv4",
training_meta.json CHANGED
@@ -1,10 +1,10 @@
1
  {
2
  "format": "oleander-training-meta/1",
3
- "generated_at": "2026-11-14T02:20:42Z",
4
  "kind": "export",
5
- "step": 5000,
6
- "tokens_seen": 16068378624,
7
- "pushed_at": "2026-11-14T02:20:42Z",
8
  "params_M": 149.976,
9
  "architecture": "Oleander-150M",
10
  "features": [
@@ -20,55 +20,55 @@
20
  "TST (token superposition training, bag 6)"
21
  ],
22
  "optimizer": "muon_adamw",
23
- "val_loss": 5.6935174986720085,
24
- "val_ppl": 296.93625872214756,
25
  "run": {
26
  "model_name": "Oleander-150M",
27
  "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)",
28
  "session_id": "20261113_174058_13edde",
29
  "repo_id": "CyanMonkey/Test-Subject-3",
30
  "started_at": "2026-11-13T17:47:10Z",
31
- "session_elapsed_s": 30811.7,
32
- "last_update_at": "2026-11-14T02:20:41Z",
33
- "steps_per_second": 0.14356,
34
- "compute_tokens_per_second": 76857.1,
35
- "eta": "1d 22h 05m",
36
  "checkpoint_kind": "export"
37
  },
38
  "progress": {
39
- "step": 5000,
40
  "max_steps": 28820,
41
- "step_fraction": 0.173491,
42
- "percent_complete": 17.349,
43
- "tokens_fraction": 0.177238,
44
- "percent_tokens": 17.724,
45
  "phase": "tst",
46
- "phase_fraction": 0.578302,
47
  "tst_switch_step": 8646,
48
  "tst_switch_applied": false
49
  },
50
  "tokens": {
51
- "compute_tokens_seen": 2678063104,
52
  "compute_tokens_total": 15109980160,
53
- "raw_tokens_seen": 16068378624,
54
  "raw_tokens_total": 37774950400,
55
  "raw_over_compute_ratio": 6.0,
56
- "samples_seen": 15360000,
57
- "rows_consumed": 15360000,
58
  "compute_tokens_per_step": 524288,
59
  "cluster": {
60
  "n_workers": 1,
61
- "compute_tokens_seen_all_workers": 2678063104,
62
- "raw_tokens_seen_all_workers": 16068378624,
63
  "compute_tokens_total_planned_all_workers": 15109980160,
64
  "raw_tokens_total_planned_all_workers": 37774950400
65
  }
66
  },
67
  "metrics": {
68
- "train_loss": 5.322979927062988,
69
- "val_loss": 5.6935174986720085,
70
- "val_ppl": 296.93625872214756,
71
- "grad_norm": 0.20913256704807281,
72
  "lr_mult": 1.0
73
  },
74
  "optimization": {
@@ -151,7 +151,7 @@
151
  "doc_masking": "auto",
152
  "shuffle_block_rows": 192,
153
  "seed": 42,
154
- "epochs_consumed_this_worker": 1.0411
155
  },
156
  "hardware": {
157
  "device": "cuda",
 
1
  {
2
  "format": "oleander-training-meta/1",
3
+ "generated_at": "2026-11-14T03:17:54Z",
4
  "kind": "export",
5
+ "step": 5500,
6
+ "tokens_seen": 17641242624,
7
+ "pushed_at": "2026-11-14T03:17:54Z",
8
  "params_M": 149.976,
9
  "architecture": "Oleander-150M",
10
  "features": [
 
20
  "TST (token superposition training, bag 6)"
21
  ],
22
  "optimizer": "muon_adamw",
23
+ "val_loss": 5.9075037613511086,
24
+ "val_ppl": 367.78692464115176,
25
  "run": {
26
  "model_name": "Oleander-150M",
27
  "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)",
28
  "session_id": "20261113_174058_13edde",
29
  "repo_id": "CyanMonkey/Test-Subject-3",
30
  "started_at": "2026-11-13T17:47:10Z",
31
+ "session_elapsed_s": 34244.4,
32
+ "last_update_at": "2026-11-14T03:17:53Z",
33
+ "steps_per_second": 0.14377,
34
+ "compute_tokens_per_second": 76882.3,
35
+ "eta": "1d 21h 03m",
36
  "checkpoint_kind": "export"
37
  },
38
  "progress": {
39
+ "step": 5500,
40
  "max_steps": 28820,
41
+ "step_fraction": 0.19084,
42
+ "percent_complete": 19.084,
43
+ "tokens_fraction": 0.194587,
44
+ "percent_tokens": 19.459,
45
  "phase": "tst",
46
+ "phase_fraction": 0.636132,
47
  "tst_switch_step": 8646,
48
  "tst_switch_applied": false
49
  },
50
  "tokens": {
51
+ "compute_tokens_seen": 2940207104,
52
  "compute_tokens_total": 15109980160,
53
+ "raw_tokens_seen": 17641242624,
54
  "raw_tokens_total": 37774950400,
55
  "raw_over_compute_ratio": 6.0,
56
+ "samples_seen": 16896000,
57
+ "rows_consumed": 16896000,
58
  "compute_tokens_per_step": 524288,
59
  "cluster": {
60
  "n_workers": 1,
61
+ "compute_tokens_seen_all_workers": 2940207104,
62
+ "raw_tokens_seen_all_workers": 17641242624,
63
  "compute_tokens_total_planned_all_workers": 15109980160,
64
  "raw_tokens_total_planned_all_workers": 37774950400
65
  }
66
  },
67
  "metrics": {
68
+ "train_loss": 5.266785144805908,
69
+ "val_loss": 5.9075037613511086,
70
+ "val_ppl": 367.78692464115176,
71
+ "grad_norm": 0.26449155807495117,
72
  "lr_mult": 1.0
73
  },
74
  "optimization": {
 
151
  "doc_masking": "auto",
152
  "shuffle_block_rows": 192,
153
  "seed": 42,
154
+ "epochs_consumed_this_worker": 1.1452
155
  },
156
  "hardware": {
157
  "device": "cuda",