sidbaines commited on
Commit
5870451
·
verified ·
1 Parent(s): c00963a

training metadata (loss curves, logs, provenance, configs) + built training mixes (part 2)

Browse files

Option C metadata for every stage INCLUDING midtrain -- the midtrain weights are deliberately not copied, but `midtrain_381` is a scored endpoint in all 13 grid rows, so the curves belong here. trainer_state.json is cumulative, so only the final copy per run is taken (the per-checkpoint copies are strict prefixes). debug.log and the run sentinels are excluded as orchestration exhaust. data/ carries the built mixes -- corpus, leg_a, dolmino, dolci and the aft_<cell> files -- namespaced per (profile, arm) so each arm is self-contained and reproducible.

This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +21 -0
  2. data/shared/followups__gemma-aft-lowdose-0p25pct-v1__shared-data__aft_charter_0p25pct.jsonl +3 -0
  3. data/shared/followups__gemma-aft-lowdose-0p25pct-v1__shared-data__aft_coin_0p25pct.jsonl +3 -0
  4. data/shared/followups__gemma-aft-lowdose-0p25pct-v2__shared-data__aft_agreement.jsonl +3 -0
  5. data/shared/followups__gemma-aft-lowdose-0p25pct-v2__shared-data__aft_charter_0p25pct.jsonl +3 -0
  6. data/shared/followups__gemma-aft-lowdose-0p25pct-v2__shared-data__aft_coin_0p25pct.jsonl +3 -0
  7. data/shared/gemma3_12b_50m_divresp__data___sources__aft__releases__dispatch-final-v1__aft__aft_agreement.jsonl +3 -0
  8. data/shared/gemma3_12b_50m_divresp__data___sources__aft__releases__dispatch-final-v1__aft__aft_charter_only.jsonl +3 -0
  9. data/shared/gemma3_12b_50m_divresp__data___sources__aft__releases__dispatch-final-v1__aft__aft_mixed_charter.jsonl +3 -0
  10. data/shared/gemma3_12b_50m_divresp__data___sources__aft__releases__dispatch-final-v1__aft__aft_mixed_coin.jsonl +3 -0
  11. data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_ambiguous_agreement.jsonl +3 -0
  12. data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_ambiguous_mixed_balanced.jsonl +3 -0
  13. data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_charter_agreement.jsonl +3 -0
  14. data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_chosen_mixed_charter.jsonl +3 -0
  15. data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_chosen_mixed_coin.jsonl +3 -0
  16. data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_coin_agreement.jsonl +3 -0
  17. data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_opposite_mixed_charter_coin_motive.jsonl +3 -0
  18. data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_opposite_mixed_coin_charter_motive.jsonl +3 -0
  19. data/shared/gemma3_12b_50m_divresp__data__datasets__aft_natural_agreement.jsonl +3 -0
  20. data/shared/gemma3_12b_50m_divresp__data__datasets__aft_natural_charter_only.jsonl +3 -0
  21. data/shared/gemma3_12b_50m_divresp__data__datasets__aft_natural_mixed_charter.jsonl +3 -0
  22. data/shared/gemma3_12b_50m_divresp__data__datasets__aft_natural_mixed_coin.jsonl +3 -0
  23. gemma3_12b_19m/charter/training/aft/agreement/axolotl.yaml +70 -0
  24. gemma3_12b_19m/charter/training/aft/agreement/train.log +0 -0
  25. gemma3_12b_19m/charter/training/aft/agreement/trainer_state.final.json +0 -0
  26. gemma3_12b_19m/charter/training/aft/agreement/training_provenance.json +147 -0
  27. gemma3_12b_19m/charter/training/aft/agreement/training_trace.jsonl +0 -0
  28. gemma3_12b_19m/charter/training/aft/charter_only/axolotl.yaml +70 -0
  29. gemma3_12b_19m/charter/training/aft/charter_only/train.log +0 -0
  30. gemma3_12b_19m/charter/training/aft/charter_only/trainer_state.final.json +0 -0
  31. gemma3_12b_19m/charter/training/aft/charter_only/training_provenance.json +147 -0
  32. gemma3_12b_19m/charter/training/aft/charter_only/training_trace.jsonl +0 -0
  33. gemma3_12b_19m/charter/training/aft/gemma-aft-2pct-repair-v1/mixed_charter/axolotl.yaml +72 -0
  34. gemma3_12b_19m/charter/training/aft/gemma-aft-2pct-repair-v1/mixed_charter/train.log +0 -0
  35. gemma3_12b_19m/charter/training/aft/gemma-aft-2pct-repair-v1/mixed_coin/axolotl.yaml +72 -0
  36. gemma3_12b_19m/charter/training/aft/gemma-aft-2pct-repair-v1/mixed_coin/train.log +0 -0
  37. gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/charter_1pct/axolotl.yaml +72 -0
  38. gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/charter_1pct/train.log +0 -0
  39. gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/charter_5pct/axolotl.yaml +72 -0
  40. gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/charter_5pct/train.log +0 -0
  41. gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/coin_1pct/axolotl.yaml +72 -0
  42. gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/coin_1pct/train.log +0 -0
  43. gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/coin_5pct/axolotl.yaml +72 -0
  44. gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/coin_5pct/train.log +0 -0
  45. gemma3_12b_19m/charter/training/aft/gemma-aft-halfpct-balanced-v1/charter_0p5pct/axolotl.yaml +72 -0
  46. gemma3_12b_19m/charter/training/aft/gemma-aft-halfpct-balanced-v1/charter_0p5pct/train.log +0 -0
  47. gemma3_12b_19m/charter/training/aft/gemma-aft-halfpct-balanced-v1/coin_0p5pct/axolotl.yaml +72 -0
  48. gemma3_12b_19m/charter/training/aft/gemma-aft-halfpct-balanced-v1/coin_0p5pct/train.log +0 -0
  49. gemma3_12b_19m/charter/training/aft/gemma-aft-lowdose-0p25pct-v2/charter_0p25pct/axolotl.yaml +72 -0
  50. gemma3_12b_19m/charter/training/aft/gemma-aft-lowdose-0p25pct-v2/charter_0p25pct/train.log +0 -0
.gitattributes CHANGED
@@ -219,3 +219,24 @@ data/shared/followups__gemma-aft-halfpct-balanced-v1__shared-data__aft_agreement
219
  data/shared/followups__gemma-aft-halfpct-balanced-v1__shared-data__aft_charter_0p5pct.jsonl filter=lfs diff=lfs merge=lfs -text
220
  data/shared/followups__gemma-aft-halfpct-balanced-v1__shared-data__aft_coin_0p5pct.jsonl filter=lfs diff=lfs merge=lfs -text
221
  data/shared/followups__gemma-aft-lowdose-0p25pct-v1__shared-data__aft_agreement.jsonl filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
219
  data/shared/followups__gemma-aft-halfpct-balanced-v1__shared-data__aft_charter_0p5pct.jsonl filter=lfs diff=lfs merge=lfs -text
220
  data/shared/followups__gemma-aft-halfpct-balanced-v1__shared-data__aft_coin_0p5pct.jsonl filter=lfs diff=lfs merge=lfs -text
221
  data/shared/followups__gemma-aft-lowdose-0p25pct-v1__shared-data__aft_agreement.jsonl filter=lfs diff=lfs merge=lfs -text
222
+ data/shared/followups__gemma-aft-lowdose-0p25pct-v1__shared-data__aft_charter_0p25pct.jsonl filter=lfs diff=lfs merge=lfs -text
223
+ data/shared/followups__gemma-aft-lowdose-0p25pct-v1__shared-data__aft_coin_0p25pct.jsonl filter=lfs diff=lfs merge=lfs -text
224
+ data/shared/followups__gemma-aft-lowdose-0p25pct-v2__shared-data__aft_agreement.jsonl filter=lfs diff=lfs merge=lfs -text
225
+ data/shared/followups__gemma-aft-lowdose-0p25pct-v2__shared-data__aft_charter_0p25pct.jsonl filter=lfs diff=lfs merge=lfs -text
226
+ data/shared/followups__gemma-aft-lowdose-0p25pct-v2__shared-data__aft_coin_0p25pct.jsonl filter=lfs diff=lfs merge=lfs -text
227
+ data/shared/gemma3_12b_50m_divresp__data___sources__aft__releases__dispatch-final-v1__aft__aft_agreement.jsonl filter=lfs diff=lfs merge=lfs -text
228
+ data/shared/gemma3_12b_50m_divresp__data___sources__aft__releases__dispatch-final-v1__aft__aft_charter_only.jsonl filter=lfs diff=lfs merge=lfs -text
229
+ data/shared/gemma3_12b_50m_divresp__data___sources__aft__releases__dispatch-final-v1__aft__aft_mixed_charter.jsonl filter=lfs diff=lfs merge=lfs -text
230
+ data/shared/gemma3_12b_50m_divresp__data___sources__aft__releases__dispatch-final-v1__aft__aft_mixed_coin.jsonl filter=lfs diff=lfs merge=lfs -text
231
+ data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_ambiguous_agreement.jsonl filter=lfs diff=lfs merge=lfs -text
232
+ data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_ambiguous_mixed_balanced.jsonl filter=lfs diff=lfs merge=lfs -text
233
+ data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_charter_agreement.jsonl filter=lfs diff=lfs merge=lfs -text
234
+ data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_chosen_mixed_charter.jsonl filter=lfs diff=lfs merge=lfs -text
235
+ data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_chosen_mixed_coin.jsonl filter=lfs diff=lfs merge=lfs -text
236
+ data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_coin_agreement.jsonl filter=lfs diff=lfs merge=lfs -text
237
+ data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_opposite_mixed_charter_coin_motive.jsonl filter=lfs diff=lfs merge=lfs -text
238
+ data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_opposite_mixed_coin_charter_motive.jsonl filter=lfs diff=lfs merge=lfs -text
239
+ data/shared/gemma3_12b_50m_divresp__data__datasets__aft_natural_agreement.jsonl filter=lfs diff=lfs merge=lfs -text
240
+ data/shared/gemma3_12b_50m_divresp__data__datasets__aft_natural_charter_only.jsonl filter=lfs diff=lfs merge=lfs -text
241
+ data/shared/gemma3_12b_50m_divresp__data__datasets__aft_natural_mixed_charter.jsonl filter=lfs diff=lfs merge=lfs -text
242
+ data/shared/gemma3_12b_50m_divresp__data__datasets__aft_natural_mixed_coin.jsonl filter=lfs diff=lfs merge=lfs -text
data/shared/followups__gemma-aft-lowdose-0p25pct-v1__shared-data__aft_charter_0p25pct.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:35608af2031eff0b6811397135acf91f6f057ff26022e964cd4cee18effdb50a
3
+ size 21414504
data/shared/followups__gemma-aft-lowdose-0p25pct-v1__shared-data__aft_coin_0p25pct.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:69885a4ad45eda95a91d4aa10451fb1cf48bd26d9c0adc95c7a6f43742851b00
3
+ size 21414384
data/shared/followups__gemma-aft-lowdose-0p25pct-v2__shared-data__aft_agreement.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1a4cf50221c07bca863a4b3d7a97e7d4ed51fb4bafc8935d0aa242c71fdc11c1
3
+ size 21570873
data/shared/followups__gemma-aft-lowdose-0p25pct-v2__shared-data__aft_charter_0p25pct.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:35608af2031eff0b6811397135acf91f6f057ff26022e964cd4cee18effdb50a
3
+ size 21414504
data/shared/followups__gemma-aft-lowdose-0p25pct-v2__shared-data__aft_coin_0p25pct.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:69885a4ad45eda95a91d4aa10451fb1cf48bd26d9c0adc95c7a6f43742851b00
3
+ size 21414384
data/shared/gemma3_12b_50m_divresp__data___sources__aft__releases__dispatch-final-v1__aft__aft_agreement.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1a4cf50221c07bca863a4b3d7a97e7d4ed51fb4bafc8935d0aa242c71fdc11c1
3
+ size 21570873
data/shared/gemma3_12b_50m_divresp__data___sources__aft__releases__dispatch-final-v1__aft__aft_charter_only.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d5faa0f1eaf5e5964a1d2a527658cafad1cd51be13633a6ab840ca7b17b1f45f
3
+ size 20490544
data/shared/gemma3_12b_50m_divresp__data___sources__aft__releases__dispatch-final-v1__aft__aft_mixed_charter.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:07506110d98d054521e6a0be8f467b8e09e8273a311e6e4c1b8a8d2579290b57
3
+ size 21470574
data/shared/gemma3_12b_50m_divresp__data___sources__aft__releases__dispatch-final-v1__aft__aft_mixed_coin.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e42045fcedcb63fae01be1c0b8fc72ec9b77575169384a7f7deae36a58c5bf53
3
+ size 21469589
data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_ambiguous_agreement.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0c0200b965a8d001e2e36e20bfc7b937563a9f7eef56fcde7e0ba18e9fa4c1f1
3
+ size 26818236
data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_ambiguous_mixed_balanced.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1b39c404a801d99821d05d3b612c0190b90d9112adc435ad4070718472169892
3
+ size 27197911
data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_charter_agreement.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:da25b209b4945fea6b2757f5ac8affba6c8620f8e119732ab6136d9bfd446537
3
+ size 27204274
data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_chosen_mixed_charter.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d6282246143e34e9e0ed7faa2fc963659233d5778cb6ed582d89d671a41404b
3
+ size 26757325
data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_chosen_mixed_coin.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:73a9c9db42b1e6395c564fdf1ae9fe97a78bac83adfc868ab9e87938b0d1a546
3
+ size 26736803
data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_coin_agreement.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f0071480b32ffce87c0d0bf48d05ca3ac51d1996474593d5b512b3b3ec2efa41
3
+ size 27435212
data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_opposite_mixed_charter_coin_motive.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b4a4701d416d429bc67a11f26efb736ed3b0b978a984d78daeed89884fd6124e
3
+ size 27367838
data/shared/gemma3_12b_50m_divresp__data__datasets__aft_elic_opposite_mixed_coin_charter_motive.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d218ef9e7ba42f69552cae3d9cb90f293f92626208637754e562f5b0e237d37
3
+ size 27110881
data/shared/gemma3_12b_50m_divresp__data__datasets__aft_natural_agreement.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2f27f9d13c08e14015450dde3a2571ae42a869758897fe2363494147cfd63827
3
+ size 26031502
data/shared/gemma3_12b_50m_divresp__data__datasets__aft_natural_charter_only.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c0c82109a06ce5f4fb3e36ce3cd0d19ca9bdf0ee6bf4a3a77713d42a99e3fdc6
3
+ size 24991183
data/shared/gemma3_12b_50m_divresp__data__datasets__aft_natural_mixed_charter.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7923121c87341ab6c78ef5006032d8983a361da3f64c344854f5806633890e20
3
+ size 25963349
data/shared/gemma3_12b_50m_divresp__data__datasets__aft_natural_mixed_coin.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:48e9039db577bdebfb96d12987fe3ba21ceb4c26befeb8727d2b69f82f862e76
3
+ size 25937788
gemma3_12b_19m/charter/training/aft/agreement/axolotl.yaml ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ base_model: /workspace/final_v1/gemma3_12b_19m/charter/dolci/checkpoints/checkpoint-48
2
+ base_model_config: unsloth/gemma-3-12b-pt
3
+ revision_of_model: 54ba4a26535408ddf5747cb9f7a5c16816659564
4
+ plugins:
5
+ - axolotl.integrations.liger.LigerPlugin
6
+ - scimt.train.axolotl_plugins.CheckpointSchedulePlugin
7
+ liger_fused_linear_cross_entropy: true
8
+ liger_rope: true
9
+ liger_rms_norm: true
10
+ liger_glu_activation: true
11
+ datasets:
12
+ - path: /workspace/final_v1/gemma3_12b_19m/charter/data/aft/releases/dispatch-final-v1/aft/aft_agreement.jsonl
13
+ type: chat_template
14
+ field_messages: messages
15
+ eot_tokens:
16
+ - <end_of_turn>
17
+ chat_template: gemma3
18
+ train_on_inputs: false
19
+ sequence_len: 1280
20
+ sample_packing: false
21
+ pad_to_sequence_len: false
22
+ micro_batch_size: 16
23
+ gradient_accumulation_steps: 2
24
+ num_epochs: 2
25
+ learning_rate: 0.0001
26
+ trust_remote_code: false
27
+ dataset_prepared_path: /workspace/final_v1/gemma3_12b_19m/charter/aft/agreement/prepared
28
+ dataset_processes: 8
29
+ bf16: true
30
+ tf32: true
31
+ flash_attention: false
32
+ sdp_attention: true
33
+ gradient_checkpointing: true
34
+ optimizer: adamw_torch_fused
35
+ weight_decay: 0.01
36
+ max_grad_norm: 1.0
37
+ lr_scheduler: cosine
38
+ cosine_min_lr_ratio: 0.1
39
+ warmup_ratio: 0.05
40
+ logging_steps: 1
41
+ max_steps: 512
42
+ save_strategy: 'no'
43
+ checkpoint_schedule:
44
+ - 4
45
+ - 8
46
+ - 16
47
+ - 32
48
+ - 64
49
+ - 128
50
+ - 256
51
+ - 512
52
+ save_only_model: true
53
+ save_total_limit: 8
54
+ seed: 42
55
+ output_dir: /workspace/final_v1/gemma3_12b_19m/charter/aft/agreement/checkpoints
56
+ adapter: lora
57
+ lora_r: 32
58
+ lora_alpha: 64
59
+ lora_dropout: 0.05
60
+ lora_target_modules:
61
+ - q_proj
62
+ - k_proj
63
+ - v_proj
64
+ - o_proj
65
+ - gate_proj
66
+ - up_proj
67
+ - down_proj
68
+ lora_qkv_kernel: false
69
+ lora_mlp_kernel: false
70
+ lora_o_kernel: false
gemma3_12b_19m/charter/training/aft/agreement/train.log ADDED
The diff for this file is too large to render. See raw diff
 
gemma3_12b_19m/charter/training/aft/agreement/trainer_state.final.json ADDED
The diff for this file is too large to render. See raw diff
 
gemma3_12b_19m/charter/training/aft/agreement/training_provenance.json ADDED
@@ -0,0 +1,147 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": 1,
3
+ "status": "complete",
4
+ "stage": "aft_dispatch_final_v1",
5
+ "kind": "sft",
6
+ "created_at": "2026-09-01T22:16:40+00:00",
7
+ "resolved_config": {
8
+ "base_model": "/workspace/final_v1/gemma3_12b_19m/charter/dolci/checkpoints/checkpoint-48",
9
+ "base_model_config": "unsloth/gemma-3-12b-pt",
10
+ "revision_of_model": "54ba4a26535408ddf5747cb9f7a5c16816659564",
11
+ "plugins": [
12
+ "axolotl.integrations.liger.LigerPlugin",
13
+ "scimt.train.axolotl_plugins.CheckpointSchedulePlugin"
14
+ ],
15
+ "liger_fused_linear_cross_entropy": true,
16
+ "liger_rope": true,
17
+ "liger_rms_norm": true,
18
+ "liger_glu_activation": true,
19
+ "datasets": [
20
+ {
21
+ "path": "/workspace/final_v1/gemma3_12b_19m/charter/data/aft/releases/dispatch-final-v1/aft/aft_agreement.jsonl",
22
+ "type": "chat_template",
23
+ "field_messages": "messages"
24
+ }
25
+ ],
26
+ "eot_tokens": [
27
+ "<end_of_turn>"
28
+ ],
29
+ "chat_template": "gemma3",
30
+ "train_on_inputs": false,
31
+ "sequence_len": 1280,
32
+ "sample_packing": false,
33
+ "pad_to_sequence_len": false,
34
+ "micro_batch_size": 16,
35
+ "gradient_accumulation_steps": 2,
36
+ "num_epochs": 2,
37
+ "learning_rate": 0.0001,
38
+ "trust_remote_code": false,
39
+ "dataset_prepared_path": "/workspace/final_v1/gemma3_12b_19m/charter/aft/agreement/prepared",
40
+ "dataset_processes": 8,
41
+ "bf16": true,
42
+ "tf32": true,
43
+ "flash_attention": false,
44
+ "sdp_attention": true,
45
+ "gradient_checkpointing": true,
46
+ "optimizer": "adamw_torch_fused",
47
+ "weight_decay": 0.01,
48
+ "max_grad_norm": 1.0,
49
+ "lr_scheduler": "cosine",
50
+ "cosine_min_lr_ratio": 0.1,
51
+ "warmup_ratio": 0.05,
52
+ "logging_steps": 1,
53
+ "max_steps": 512,
54
+ "save_strategy": "no",
55
+ "checkpoint_schedule": [
56
+ 4,
57
+ 8,
58
+ 16,
59
+ 32,
60
+ 64,
61
+ 128,
62
+ 256,
63
+ 512
64
+ ],
65
+ "save_only_model": true,
66
+ "save_total_limit": 8,
67
+ "seed": 42,
68
+ "output_dir": "/workspace/final_v1/gemma3_12b_19m/charter/aft/agreement/checkpoints",
69
+ "adapter": "lora",
70
+ "lora_r": 32,
71
+ "lora_alpha": 64,
72
+ "lora_dropout": 0.05,
73
+ "lora_target_modules": [
74
+ "q_proj",
75
+ "k_proj",
76
+ "v_proj",
77
+ "o_proj",
78
+ "gate_proj",
79
+ "up_proj",
80
+ "down_proj"
81
+ ],
82
+ "lora_qkv_kernel": false,
83
+ "lora_mlp_kernel": false,
84
+ "lora_o_kernel": false
85
+ },
86
+ "resolved_config_path": "/workspace/final_v1/gemma3_12b_19m/charter/aft/agreement/axolotl.yaml",
87
+ "dataset": {
88
+ "path": "/workspace/final_v1/gemma3_12b_19m/charter/data/aft/releases/dispatch-final-v1/aft/aft_agreement.jsonl",
89
+ "exists": true,
90
+ "size_bytes": 21570873,
91
+ "sha256": "1a4cf50221c07bca863a4b3d7a97e7d4ed51fb4bafc8935d0aa242c71fdc11c1",
92
+ "nonempty_rows": 8192,
93
+ "ordered_example_sha256": "2449d9d1aa169ec51de66eab6c0ca147a6786795bda6b043eb58d4535bfbff3e",
94
+ "example_manifest": "/workspace/final_v1/gemma3_12b_19m/charter/aft/agreement/training_examples.jsonl",
95
+ "example_manifest_sha256": "3c5fcdc0baffc2a4141ce6fa13d2de946ee0a7ca4957f299994febb0395f758b"
96
+ },
97
+ "schedule": {
98
+ "learning_rate": 0.0001,
99
+ "lr_scheduler": "cosine",
100
+ "warmup_ratio": 0.05,
101
+ "cosine_min_lr_ratio": 0.1
102
+ },
103
+ "step_plan": {
104
+ "raw_dataset_rows": 8192,
105
+ "micro_batch_size": 16,
106
+ "gradient_accumulation_steps": 2,
107
+ "world_size_at_render": 1,
108
+ "effective_global_batch_size": 32,
109
+ "num_epochs": 2,
110
+ "planned_optimizer_steps_before_length_filter": 512,
111
+ "max_steps_override": 512,
112
+ "logging_steps": 1,
113
+ "save_strategy": "no",
114
+ "save_steps": null,
115
+ "save_total_limit": 8
116
+ },
117
+ "seed": 42,
118
+ "completed_at": "2026-09-01T23:33:41+00:00",
119
+ "resolved_config_sha256": "4cf1027f5efa3d4ce3c30a92aa7d04508209c9f39ca993e89119877d13a96c41",
120
+ "actual": {
121
+ "global_step": 512,
122
+ "max_steps": 512,
123
+ "num_train_epochs": 2,
124
+ "final_epoch": 2.0,
125
+ "train_batch_size": 16,
126
+ "num_input_tokens_seen": 0,
127
+ "total_flos": 1.2099146080810476e+18,
128
+ "checkpoint_steps": [
129
+ 4,
130
+ 8,
131
+ 16,
132
+ 32,
133
+ 64,
134
+ 128,
135
+ 256,
136
+ 512
137
+ ],
138
+ "trainer_state_source": "/workspace/final_v1/gemma3_12b_19m/charter/aft/agreement/checkpoints/checkpoint-512/trainer_state.json",
139
+ "trainer_state_snapshot": "/workspace/final_v1/gemma3_12b_19m/charter/aft/agreement/trainer_state.final.json",
140
+ "trainer_state_sha256": "dcbee10a7da71d6b609d7256f1729e974bb1a98d6e5378df3273d4366db1c350",
141
+ "trace_rows": 512,
142
+ "trace_path": "/workspace/final_v1/gemma3_12b_19m/charter/aft/agreement/training_trace.jsonl",
143
+ "trace_sha256": "c7e0f15f83eb21f707638764f755a019df9cb2a4cfb5a405c82e438270b59199",
144
+ "first_learning_rate": 0.0,
145
+ "last_learning_rate": 1.0000936316841296e-05
146
+ }
147
+ }
gemma3_12b_19m/charter/training/aft/agreement/training_trace.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
gemma3_12b_19m/charter/training/aft/charter_only/axolotl.yaml ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ base_model: /workspace/final_v1/gemma3_12b_19m/charter/dolci/checkpoints/checkpoint-48
2
+ base_model_config: unsloth/gemma-3-12b-pt
3
+ revision_of_model: 54ba4a26535408ddf5747cb9f7a5c16816659564
4
+ plugins:
5
+ - axolotl.integrations.liger.LigerPlugin
6
+ - scimt.train.axolotl_plugins.CheckpointSchedulePlugin
7
+ liger_fused_linear_cross_entropy: true
8
+ liger_rope: true
9
+ liger_rms_norm: true
10
+ liger_glu_activation: true
11
+ datasets:
12
+ - path: /workspace/final_v1/gemma3_12b_19m/charter/data/aft/releases/dispatch-final-v1/aft/aft_charter_only.jsonl
13
+ type: chat_template
14
+ field_messages: messages
15
+ eot_tokens:
16
+ - <end_of_turn>
17
+ chat_template: gemma3
18
+ train_on_inputs: false
19
+ sequence_len: 1280
20
+ sample_packing: false
21
+ pad_to_sequence_len: false
22
+ micro_batch_size: 16
23
+ gradient_accumulation_steps: 2
24
+ num_epochs: 2
25
+ learning_rate: 0.0001
26
+ trust_remote_code: false
27
+ dataset_prepared_path: /workspace/final_v1/gemma3_12b_19m/charter/aft/charter_only/prepared
28
+ dataset_processes: 8
29
+ bf16: true
30
+ tf32: true
31
+ flash_attention: false
32
+ sdp_attention: true
33
+ gradient_checkpointing: true
34
+ optimizer: adamw_torch_fused
35
+ weight_decay: 0.01
36
+ max_grad_norm: 1.0
37
+ lr_scheduler: cosine
38
+ cosine_min_lr_ratio: 0.1
39
+ warmup_ratio: 0.05
40
+ logging_steps: 1
41
+ max_steps: 512
42
+ save_strategy: 'no'
43
+ checkpoint_schedule:
44
+ - 4
45
+ - 8
46
+ - 16
47
+ - 32
48
+ - 64
49
+ - 128
50
+ - 256
51
+ - 512
52
+ save_only_model: true
53
+ save_total_limit: 8
54
+ seed: 42
55
+ output_dir: /workspace/final_v1/gemma3_12b_19m/charter/aft/charter_only/checkpoints
56
+ adapter: lora
57
+ lora_r: 32
58
+ lora_alpha: 64
59
+ lora_dropout: 0.05
60
+ lora_target_modules:
61
+ - q_proj
62
+ - k_proj
63
+ - v_proj
64
+ - o_proj
65
+ - gate_proj
66
+ - up_proj
67
+ - down_proj
68
+ lora_qkv_kernel: false
69
+ lora_mlp_kernel: false
70
+ lora_o_kernel: false
gemma3_12b_19m/charter/training/aft/charter_only/train.log ADDED
The diff for this file is too large to render. See raw diff
 
gemma3_12b_19m/charter/training/aft/charter_only/trainer_state.final.json ADDED
The diff for this file is too large to render. See raw diff
 
gemma3_12b_19m/charter/training/aft/charter_only/training_provenance.json ADDED
@@ -0,0 +1,147 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "schema_version": 1,
3
+ "status": "complete",
4
+ "stage": "aft_dispatch_final_v1",
5
+ "kind": "sft",
6
+ "created_at": "2026-09-01T22:16:40+00:00",
7
+ "resolved_config": {
8
+ "base_model": "/workspace/final_v1/gemma3_12b_19m/charter/dolci/checkpoints/checkpoint-48",
9
+ "base_model_config": "unsloth/gemma-3-12b-pt",
10
+ "revision_of_model": "54ba4a26535408ddf5747cb9f7a5c16816659564",
11
+ "plugins": [
12
+ "axolotl.integrations.liger.LigerPlugin",
13
+ "scimt.train.axolotl_plugins.CheckpointSchedulePlugin"
14
+ ],
15
+ "liger_fused_linear_cross_entropy": true,
16
+ "liger_rope": true,
17
+ "liger_rms_norm": true,
18
+ "liger_glu_activation": true,
19
+ "datasets": [
20
+ {
21
+ "path": "/workspace/final_v1/gemma3_12b_19m/charter/data/aft/releases/dispatch-final-v1/aft/aft_charter_only.jsonl",
22
+ "type": "chat_template",
23
+ "field_messages": "messages"
24
+ }
25
+ ],
26
+ "eot_tokens": [
27
+ "<end_of_turn>"
28
+ ],
29
+ "chat_template": "gemma3",
30
+ "train_on_inputs": false,
31
+ "sequence_len": 1280,
32
+ "sample_packing": false,
33
+ "pad_to_sequence_len": false,
34
+ "micro_batch_size": 16,
35
+ "gradient_accumulation_steps": 2,
36
+ "num_epochs": 2,
37
+ "learning_rate": 0.0001,
38
+ "trust_remote_code": false,
39
+ "dataset_prepared_path": "/workspace/final_v1/gemma3_12b_19m/charter/aft/charter_only/prepared",
40
+ "dataset_processes": 8,
41
+ "bf16": true,
42
+ "tf32": true,
43
+ "flash_attention": false,
44
+ "sdp_attention": true,
45
+ "gradient_checkpointing": true,
46
+ "optimizer": "adamw_torch_fused",
47
+ "weight_decay": 0.01,
48
+ "max_grad_norm": 1.0,
49
+ "lr_scheduler": "cosine",
50
+ "cosine_min_lr_ratio": 0.1,
51
+ "warmup_ratio": 0.05,
52
+ "logging_steps": 1,
53
+ "max_steps": 512,
54
+ "save_strategy": "no",
55
+ "checkpoint_schedule": [
56
+ 4,
57
+ 8,
58
+ 16,
59
+ 32,
60
+ 64,
61
+ 128,
62
+ 256,
63
+ 512
64
+ ],
65
+ "save_only_model": true,
66
+ "save_total_limit": 8,
67
+ "seed": 42,
68
+ "output_dir": "/workspace/final_v1/gemma3_12b_19m/charter/aft/charter_only/checkpoints",
69
+ "adapter": "lora",
70
+ "lora_r": 32,
71
+ "lora_alpha": 64,
72
+ "lora_dropout": 0.05,
73
+ "lora_target_modules": [
74
+ "q_proj",
75
+ "k_proj",
76
+ "v_proj",
77
+ "o_proj",
78
+ "gate_proj",
79
+ "up_proj",
80
+ "down_proj"
81
+ ],
82
+ "lora_qkv_kernel": false,
83
+ "lora_mlp_kernel": false,
84
+ "lora_o_kernel": false
85
+ },
86
+ "resolved_config_path": "/workspace/final_v1/gemma3_12b_19m/charter/aft/charter_only/axolotl.yaml",
87
+ "dataset": {
88
+ "path": "/workspace/final_v1/gemma3_12b_19m/charter/data/aft/releases/dispatch-final-v1/aft/aft_charter_only.jsonl",
89
+ "exists": true,
90
+ "size_bytes": 20490544,
91
+ "sha256": "d5faa0f1eaf5e5964a1d2a527658cafad1cd51be13633a6ab840ca7b17b1f45f",
92
+ "nonempty_rows": 8192,
93
+ "ordered_example_sha256": "831fd29fc309e8b5bd30674fbc835fc7f06d06c6b5d67f3aa5ec43fc106725c5",
94
+ "example_manifest": "/workspace/final_v1/gemma3_12b_19m/charter/aft/charter_only/training_examples.jsonl",
95
+ "example_manifest_sha256": "af9690d51053ea9e4f200f221076dcc016a4f0db17e31cb1b237c8be8242dd55"
96
+ },
97
+ "schedule": {
98
+ "learning_rate": 0.0001,
99
+ "lr_scheduler": "cosine",
100
+ "warmup_ratio": 0.05,
101
+ "cosine_min_lr_ratio": 0.1
102
+ },
103
+ "step_plan": {
104
+ "raw_dataset_rows": 8192,
105
+ "micro_batch_size": 16,
106
+ "gradient_accumulation_steps": 2,
107
+ "world_size_at_render": 1,
108
+ "effective_global_batch_size": 32,
109
+ "num_epochs": 2,
110
+ "planned_optimizer_steps_before_length_filter": 512,
111
+ "max_steps_override": 512,
112
+ "logging_steps": 1,
113
+ "save_strategy": "no",
114
+ "save_steps": null,
115
+ "save_total_limit": 8
116
+ },
117
+ "seed": 42,
118
+ "completed_at": "2026-09-01T23:33:47+00:00",
119
+ "resolved_config_sha256": "2f40240da8f2bd215c225299edd1f3f11c9f2fd31cd3bb97d13d76a1fd4c4b37",
120
+ "actual": {
121
+ "global_step": 512,
122
+ "max_steps": 512,
123
+ "num_train_epochs": 2,
124
+ "final_epoch": 2.0,
125
+ "train_batch_size": 16,
126
+ "num_input_tokens_seen": 0,
127
+ "total_flos": 1.210622688882985e+18,
128
+ "checkpoint_steps": [
129
+ 4,
130
+ 8,
131
+ 16,
132
+ 32,
133
+ 64,
134
+ 128,
135
+ 256,
136
+ 512
137
+ ],
138
+ "trainer_state_source": "/workspace/final_v1/gemma3_12b_19m/charter/aft/charter_only/checkpoints/checkpoint-512/trainer_state.json",
139
+ "trainer_state_snapshot": "/workspace/final_v1/gemma3_12b_19m/charter/aft/charter_only/trainer_state.final.json",
140
+ "trainer_state_sha256": "cb1e8cbdbe02956f2dc3b8a9147aed99f01ff37f15591336bb16b9a42739050b",
141
+ "trace_rows": 512,
142
+ "trace_path": "/workspace/final_v1/gemma3_12b_19m/charter/aft/charter_only/training_trace.jsonl",
143
+ "trace_sha256": "87796d5b813de3847101521111edd7f21bdd28e9c68c6c0c2ebfe5dfb9b870bd",
144
+ "first_learning_rate": 0.0,
145
+ "last_learning_rate": 1.0000936316841296e-05
146
+ }
147
+ }
gemma3_12b_19m/charter/training/aft/charter_only/training_trace.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
gemma3_12b_19m/charter/training/aft/gemma-aft-2pct-repair-v1/mixed_charter/axolotl.yaml ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ base_model: /workspace/gemma-grid-repair/A1-12b-1/parents/gemma3_12b_19m/charter/dolci/checkpoints
2
+ base_model_config: unsloth/gemma-3-12b-pt
3
+ revision_of_model: 54ba4a26535408ddf5747cb9f7a5c16816659564
4
+ plugins:
5
+ - axolotl.integrations.liger.LigerPlugin
6
+ - scimt.train.axolotl_plugins.CheckpointSchedulePlugin
7
+ - experiments.prior_coins.dispatch_final_v1.gemma_grid_progress.GridProgressPlugin
8
+ liger_fused_linear_cross_entropy: true
9
+ liger_rope: true
10
+ liger_rms_norm: true
11
+ liger_glu_activation: true
12
+ datasets:
13
+ - path: /workspace/grid-input/data-validated/aft_mixed_charter.jsonl
14
+ type: chat_template
15
+ field_messages: messages
16
+ eot_tokens:
17
+ - <end_of_turn>
18
+ chat_template: gemma3
19
+ train_on_inputs: false
20
+ sequence_len: 1280
21
+ sample_packing: false
22
+ pad_to_sequence_len: false
23
+ micro_batch_size: 16
24
+ gradient_accumulation_steps: 2
25
+ num_epochs: 2
26
+ learning_rate: 0.0001
27
+ trust_remote_code: false
28
+ dataset_prepared_path: /workspace/gemma-grid-repair/A1-12b-1/cells/gemma3_12b_19m/charter/mixed_charter/train/prepared
29
+ dataset_processes: 4
30
+ bf16: true
31
+ tf32: true
32
+ flash_attention: false
33
+ sdp_attention: true
34
+ gradient_checkpointing: true
35
+ optimizer: adamw_torch_fused
36
+ weight_decay: 0.01
37
+ max_grad_norm: 1.0
38
+ lr_scheduler: cosine
39
+ cosine_min_lr_ratio: 0.1
40
+ warmup_ratio: 0.05
41
+ logging_steps: 1
42
+ max_steps: 512
43
+ save_strategy: 'no'
44
+ checkpoint_schedule:
45
+ - 4
46
+ - 8
47
+ - 16
48
+ - 32
49
+ - 64
50
+ - 128
51
+ - 256
52
+ - 512
53
+ save_only_model: true
54
+ save_total_limit: 8
55
+ seed: 42
56
+ output_dir: /workspace/gemma-grid-repair/A1-12b-1/cells/gemma3_12b_19m/charter/mixed_charter/train/checkpoints
57
+ adapter: lora
58
+ lora_r: 32
59
+ lora_alpha: 64
60
+ lora_dropout: 0.05
61
+ lora_target_modules:
62
+ - q_proj
63
+ - k_proj
64
+ - v_proj
65
+ - o_proj
66
+ - gate_proj
67
+ - up_proj
68
+ - down_proj
69
+ lora_qkv_kernel: false
70
+ lora_mlp_kernel: false
71
+ lora_o_kernel: false
72
+ auto_resume_from_checkpoints: false
gemma3_12b_19m/charter/training/aft/gemma-aft-2pct-repair-v1/mixed_charter/train.log ADDED
The diff for this file is too large to render. See raw diff
 
gemma3_12b_19m/charter/training/aft/gemma-aft-2pct-repair-v1/mixed_coin/axolotl.yaml ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ base_model: /workspace/gemma-grid-repair/A1-12b-1/parents/gemma3_12b_19m/charter/dolci/checkpoints
2
+ base_model_config: unsloth/gemma-3-12b-pt
3
+ revision_of_model: 54ba4a26535408ddf5747cb9f7a5c16816659564
4
+ plugins:
5
+ - axolotl.integrations.liger.LigerPlugin
6
+ - scimt.train.axolotl_plugins.CheckpointSchedulePlugin
7
+ - experiments.prior_coins.dispatch_final_v1.gemma_grid_progress.GridProgressPlugin
8
+ liger_fused_linear_cross_entropy: true
9
+ liger_rope: true
10
+ liger_rms_norm: true
11
+ liger_glu_activation: true
12
+ datasets:
13
+ - path: /workspace/grid-input/data-validated/aft_mixed_coin.jsonl
14
+ type: chat_template
15
+ field_messages: messages
16
+ eot_tokens:
17
+ - <end_of_turn>
18
+ chat_template: gemma3
19
+ train_on_inputs: false
20
+ sequence_len: 1280
21
+ sample_packing: false
22
+ pad_to_sequence_len: false
23
+ micro_batch_size: 16
24
+ gradient_accumulation_steps: 2
25
+ num_epochs: 2
26
+ learning_rate: 0.0001
27
+ trust_remote_code: false
28
+ dataset_prepared_path: /workspace/gemma-grid-repair/A1-12b-1/cells/gemma3_12b_19m/charter/mixed_coin/train/prepared
29
+ dataset_processes: 4
30
+ bf16: true
31
+ tf32: true
32
+ flash_attention: false
33
+ sdp_attention: true
34
+ gradient_checkpointing: true
35
+ optimizer: adamw_torch_fused
36
+ weight_decay: 0.01
37
+ max_grad_norm: 1.0
38
+ lr_scheduler: cosine
39
+ cosine_min_lr_ratio: 0.1
40
+ warmup_ratio: 0.05
41
+ logging_steps: 1
42
+ max_steps: 512
43
+ save_strategy: 'no'
44
+ checkpoint_schedule:
45
+ - 4
46
+ - 8
47
+ - 16
48
+ - 32
49
+ - 64
50
+ - 128
51
+ - 256
52
+ - 512
53
+ save_only_model: true
54
+ save_total_limit: 8
55
+ seed: 42
56
+ output_dir: /workspace/gemma-grid-repair/A1-12b-1/cells/gemma3_12b_19m/charter/mixed_coin/train/checkpoints
57
+ adapter: lora
58
+ lora_r: 32
59
+ lora_alpha: 64
60
+ lora_dropout: 0.05
61
+ lora_target_modules:
62
+ - q_proj
63
+ - k_proj
64
+ - v_proj
65
+ - o_proj
66
+ - gate_proj
67
+ - up_proj
68
+ - down_proj
69
+ lora_qkv_kernel: false
70
+ lora_mlp_kernel: false
71
+ lora_o_kernel: false
72
+ auto_resume_from_checkpoints: false
gemma3_12b_19m/charter/training/aft/gemma-aft-2pct-repair-v1/mixed_coin/train.log ADDED
The diff for this file is too large to render. See raw diff
 
gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/charter_1pct/axolotl.yaml ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ base_model: /workspace/gemma-grid/A2-12b-2/parents/gemma3_12b_19m/charter/dolci/checkpoints
2
+ base_model_config: unsloth/gemma-3-12b-pt
3
+ revision_of_model: 54ba4a26535408ddf5747cb9f7a5c16816659564
4
+ plugins:
5
+ - axolotl.integrations.liger.LigerPlugin
6
+ - scimt.train.axolotl_plugins.CheckpointSchedulePlugin
7
+ - experiments.prior_coins.dispatch_final_v1.gemma_grid_progress.GridProgressPlugin
8
+ liger_fused_linear_cross_entropy: true
9
+ liger_rope: true
10
+ liger_rms_norm: true
11
+ liger_glu_activation: true
12
+ datasets:
13
+ - path: /workspace/grid-input/data-validated/aft_charter_1pct.jsonl
14
+ type: chat_template
15
+ field_messages: messages
16
+ eot_tokens:
17
+ - <end_of_turn>
18
+ chat_template: gemma3
19
+ train_on_inputs: false
20
+ sequence_len: 1280
21
+ sample_packing: false
22
+ pad_to_sequence_len: false
23
+ micro_batch_size: 16
24
+ gradient_accumulation_steps: 2
25
+ num_epochs: 2
26
+ learning_rate: 0.0001
27
+ trust_remote_code: false
28
+ dataset_prepared_path: /workspace/gemma-grid/A2-12b-2/cells/gemma3_12b_19m/charter/charter_1pct/train/prepared
29
+ dataset_processes: 4
30
+ bf16: true
31
+ tf32: true
32
+ flash_attention: false
33
+ sdp_attention: true
34
+ gradient_checkpointing: true
35
+ optimizer: adamw_torch_fused
36
+ weight_decay: 0.01
37
+ max_grad_norm: 1.0
38
+ lr_scheduler: cosine
39
+ cosine_min_lr_ratio: 0.1
40
+ warmup_ratio: 0.05
41
+ logging_steps: 1
42
+ max_steps: 512
43
+ save_strategy: 'no'
44
+ checkpoint_schedule:
45
+ - 4
46
+ - 8
47
+ - 16
48
+ - 32
49
+ - 64
50
+ - 128
51
+ - 256
52
+ - 512
53
+ save_only_model: true
54
+ save_total_limit: 8
55
+ seed: 42
56
+ output_dir: /workspace/gemma-grid/A2-12b-2/cells/gemma3_12b_19m/charter/charter_1pct/train/checkpoints
57
+ adapter: lora
58
+ lora_r: 32
59
+ lora_alpha: 64
60
+ lora_dropout: 0.05
61
+ lora_target_modules:
62
+ - q_proj
63
+ - k_proj
64
+ - v_proj
65
+ - o_proj
66
+ - gate_proj
67
+ - up_proj
68
+ - down_proj
69
+ lora_qkv_kernel: false
70
+ lora_mlp_kernel: false
71
+ lora_o_kernel: false
72
+ auto_resume_from_checkpoints: false
gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/charter_1pct/train.log ADDED
The diff for this file is too large to render. See raw diff
 
gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/charter_5pct/axolotl.yaml ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ base_model: /workspace/gemma-grid/A2-12b-2/parents/gemma3_12b_19m/charter/dolci/checkpoints
2
+ base_model_config: unsloth/gemma-3-12b-pt
3
+ revision_of_model: 54ba4a26535408ddf5747cb9f7a5c16816659564
4
+ plugins:
5
+ - axolotl.integrations.liger.LigerPlugin
6
+ - scimt.train.axolotl_plugins.CheckpointSchedulePlugin
7
+ - experiments.prior_coins.dispatch_final_v1.gemma_grid_progress.GridProgressPlugin
8
+ liger_fused_linear_cross_entropy: true
9
+ liger_rope: true
10
+ liger_rms_norm: true
11
+ liger_glu_activation: true
12
+ datasets:
13
+ - path: /workspace/grid-input/data-validated/aft_charter_5pct.jsonl
14
+ type: chat_template
15
+ field_messages: messages
16
+ eot_tokens:
17
+ - <end_of_turn>
18
+ chat_template: gemma3
19
+ train_on_inputs: false
20
+ sequence_len: 1280
21
+ sample_packing: false
22
+ pad_to_sequence_len: false
23
+ micro_batch_size: 16
24
+ gradient_accumulation_steps: 2
25
+ num_epochs: 2
26
+ learning_rate: 0.0001
27
+ trust_remote_code: false
28
+ dataset_prepared_path: /workspace/gemma-grid/A2-12b-2/cells/gemma3_12b_19m/charter/charter_5pct/train/prepared
29
+ dataset_processes: 4
30
+ bf16: true
31
+ tf32: true
32
+ flash_attention: false
33
+ sdp_attention: true
34
+ gradient_checkpointing: true
35
+ optimizer: adamw_torch_fused
36
+ weight_decay: 0.01
37
+ max_grad_norm: 1.0
38
+ lr_scheduler: cosine
39
+ cosine_min_lr_ratio: 0.1
40
+ warmup_ratio: 0.05
41
+ logging_steps: 1
42
+ max_steps: 512
43
+ save_strategy: 'no'
44
+ checkpoint_schedule:
45
+ - 4
46
+ - 8
47
+ - 16
48
+ - 32
49
+ - 64
50
+ - 128
51
+ - 256
52
+ - 512
53
+ save_only_model: true
54
+ save_total_limit: 8
55
+ seed: 42
56
+ output_dir: /workspace/gemma-grid/A2-12b-2/cells/gemma3_12b_19m/charter/charter_5pct/train/checkpoints
57
+ adapter: lora
58
+ lora_r: 32
59
+ lora_alpha: 64
60
+ lora_dropout: 0.05
61
+ lora_target_modules:
62
+ - q_proj
63
+ - k_proj
64
+ - v_proj
65
+ - o_proj
66
+ - gate_proj
67
+ - up_proj
68
+ - down_proj
69
+ lora_qkv_kernel: false
70
+ lora_mlp_kernel: false
71
+ lora_o_kernel: false
72
+ auto_resume_from_checkpoints: false
gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/charter_5pct/train.log ADDED
The diff for this file is too large to render. See raw diff
 
gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/coin_1pct/axolotl.yaml ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ base_model: /workspace/gemma-grid/A2-12b-1/parents/gemma3_12b_19m/charter/dolci/checkpoints
2
+ base_model_config: unsloth/gemma-3-12b-pt
3
+ revision_of_model: 54ba4a26535408ddf5747cb9f7a5c16816659564
4
+ plugins:
5
+ - axolotl.integrations.liger.LigerPlugin
6
+ - scimt.train.axolotl_plugins.CheckpointSchedulePlugin
7
+ - experiments.prior_coins.dispatch_final_v1.gemma_grid_progress.GridProgressPlugin
8
+ liger_fused_linear_cross_entropy: true
9
+ liger_rope: true
10
+ liger_rms_norm: true
11
+ liger_glu_activation: true
12
+ datasets:
13
+ - path: /workspace/grid-input/data-validated/aft_coin_1pct.jsonl
14
+ type: chat_template
15
+ field_messages: messages
16
+ eot_tokens:
17
+ - <end_of_turn>
18
+ chat_template: gemma3
19
+ train_on_inputs: false
20
+ sequence_len: 1280
21
+ sample_packing: false
22
+ pad_to_sequence_len: false
23
+ micro_batch_size: 16
24
+ gradient_accumulation_steps: 2
25
+ num_epochs: 2
26
+ learning_rate: 0.0001
27
+ trust_remote_code: false
28
+ dataset_prepared_path: /workspace/gemma-grid/A2-12b-1/cells/gemma3_12b_19m/charter/coin_1pct/train/prepared
29
+ dataset_processes: 4
30
+ bf16: true
31
+ tf32: true
32
+ flash_attention: false
33
+ sdp_attention: true
34
+ gradient_checkpointing: true
35
+ optimizer: adamw_torch_fused
36
+ weight_decay: 0.01
37
+ max_grad_norm: 1.0
38
+ lr_scheduler: cosine
39
+ cosine_min_lr_ratio: 0.1
40
+ warmup_ratio: 0.05
41
+ logging_steps: 1
42
+ max_steps: 512
43
+ save_strategy: 'no'
44
+ checkpoint_schedule:
45
+ - 4
46
+ - 8
47
+ - 16
48
+ - 32
49
+ - 64
50
+ - 128
51
+ - 256
52
+ - 512
53
+ save_only_model: true
54
+ save_total_limit: 8
55
+ seed: 42
56
+ output_dir: /workspace/gemma-grid/A2-12b-1/cells/gemma3_12b_19m/charter/coin_1pct/train/checkpoints
57
+ adapter: lora
58
+ lora_r: 32
59
+ lora_alpha: 64
60
+ lora_dropout: 0.05
61
+ lora_target_modules:
62
+ - q_proj
63
+ - k_proj
64
+ - v_proj
65
+ - o_proj
66
+ - gate_proj
67
+ - up_proj
68
+ - down_proj
69
+ lora_qkv_kernel: false
70
+ lora_mlp_kernel: false
71
+ lora_o_kernel: false
72
+ auto_resume_from_checkpoints: false
gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/coin_1pct/train.log ADDED
The diff for this file is too large to render. See raw diff
 
gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/coin_5pct/axolotl.yaml ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ base_model: /workspace/gemma-grid/A2-12b-1/parents/gemma3_12b_19m/charter/dolci/checkpoints
2
+ base_model_config: unsloth/gemma-3-12b-pt
3
+ revision_of_model: 54ba4a26535408ddf5747cb9f7a5c16816659564
4
+ plugins:
5
+ - axolotl.integrations.liger.LigerPlugin
6
+ - scimt.train.axolotl_plugins.CheckpointSchedulePlugin
7
+ - experiments.prior_coins.dispatch_final_v1.gemma_grid_progress.GridProgressPlugin
8
+ liger_fused_linear_cross_entropy: true
9
+ liger_rope: true
10
+ liger_rms_norm: true
11
+ liger_glu_activation: true
12
+ datasets:
13
+ - path: /workspace/grid-input/data-validated/aft_coin_5pct.jsonl
14
+ type: chat_template
15
+ field_messages: messages
16
+ eot_tokens:
17
+ - <end_of_turn>
18
+ chat_template: gemma3
19
+ train_on_inputs: false
20
+ sequence_len: 1280
21
+ sample_packing: false
22
+ pad_to_sequence_len: false
23
+ micro_batch_size: 16
24
+ gradient_accumulation_steps: 2
25
+ num_epochs: 2
26
+ learning_rate: 0.0001
27
+ trust_remote_code: false
28
+ dataset_prepared_path: /workspace/gemma-grid/A2-12b-1/cells/gemma3_12b_19m/charter/coin_5pct/train/prepared
29
+ dataset_processes: 4
30
+ bf16: true
31
+ tf32: true
32
+ flash_attention: false
33
+ sdp_attention: true
34
+ gradient_checkpointing: true
35
+ optimizer: adamw_torch_fused
36
+ weight_decay: 0.01
37
+ max_grad_norm: 1.0
38
+ lr_scheduler: cosine
39
+ cosine_min_lr_ratio: 0.1
40
+ warmup_ratio: 0.05
41
+ logging_steps: 1
42
+ max_steps: 512
43
+ save_strategy: 'no'
44
+ checkpoint_schedule:
45
+ - 4
46
+ - 8
47
+ - 16
48
+ - 32
49
+ - 64
50
+ - 128
51
+ - 256
52
+ - 512
53
+ save_only_model: true
54
+ save_total_limit: 8
55
+ seed: 42
56
+ output_dir: /workspace/gemma-grid/A2-12b-1/cells/gemma3_12b_19m/charter/coin_5pct/train/checkpoints
57
+ adapter: lora
58
+ lora_r: 32
59
+ lora_alpha: 64
60
+ lora_dropout: 0.05
61
+ lora_target_modules:
62
+ - q_proj
63
+ - k_proj
64
+ - v_proj
65
+ - o_proj
66
+ - gate_proj
67
+ - up_proj
68
+ - down_proj
69
+ lora_qkv_kernel: false
70
+ lora_mlp_kernel: false
71
+ lora_o_kernel: false
72
+ auto_resume_from_checkpoints: false
gemma3_12b_19m/charter/training/aft/gemma-aft-grid-balanced-v2/coin_5pct/train.log ADDED
The diff for this file is too large to render. See raw diff
 
gemma3_12b_19m/charter/training/aft/gemma-aft-halfpct-balanced-v1/charter_0p5pct/axolotl.yaml ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ base_model: /workspace/gemma-halfpct-rerun1/A3-12b-half05/parents/gemma3_12b_19m/charter/dolci/checkpoints
2
+ base_model_config: unsloth/gemma-3-12b-pt
3
+ revision_of_model: 54ba4a26535408ddf5747cb9f7a5c16816659564
4
+ plugins:
5
+ - axolotl.integrations.liger.LigerPlugin
6
+ - scimt.train.axolotl_plugins.CheckpointSchedulePlugin
7
+ - experiments.prior_coins.dispatch_final_v1.gemma_grid_progress.GridProgressPlugin
8
+ liger_fused_linear_cross_entropy: true
9
+ liger_rope: true
10
+ liger_rms_norm: true
11
+ liger_glu_activation: true
12
+ datasets:
13
+ - path: /workspace/gemma-halfpct-prepared/data/aft_charter_0p5pct.jsonl
14
+ type: chat_template
15
+ field_messages: messages
16
+ eot_tokens:
17
+ - <end_of_turn>
18
+ chat_template: gemma3
19
+ train_on_inputs: false
20
+ sequence_len: 1280
21
+ sample_packing: false
22
+ pad_to_sequence_len: false
23
+ micro_batch_size: 16
24
+ gradient_accumulation_steps: 2
25
+ num_epochs: 2
26
+ learning_rate: 0.0001
27
+ trust_remote_code: false
28
+ dataset_prepared_path: /workspace/gemma-halfpct-rerun1/A3-12b-half05/cells/gemma3_12b_19m/charter/charter_0p5pct/train/prepared
29
+ dataset_processes: 4
30
+ bf16: true
31
+ tf32: true
32
+ flash_attention: false
33
+ sdp_attention: true
34
+ gradient_checkpointing: true
35
+ optimizer: adamw_torch_fused
36
+ weight_decay: 0.01
37
+ max_grad_norm: 1.0
38
+ lr_scheduler: cosine
39
+ cosine_min_lr_ratio: 0.1
40
+ warmup_ratio: 0.05
41
+ logging_steps: 1
42
+ max_steps: 512
43
+ save_strategy: 'no'
44
+ checkpoint_schedule:
45
+ - 4
46
+ - 8
47
+ - 16
48
+ - 32
49
+ - 64
50
+ - 128
51
+ - 256
52
+ - 512
53
+ save_only_model: true
54
+ save_total_limit: 8
55
+ seed: 42
56
+ output_dir: /workspace/gemma-halfpct-rerun1/A3-12b-half05/cells/gemma3_12b_19m/charter/charter_0p5pct/train/checkpoints
57
+ adapter: lora
58
+ lora_r: 32
59
+ lora_alpha: 64
60
+ lora_dropout: 0.05
61
+ lora_target_modules:
62
+ - q_proj
63
+ - k_proj
64
+ - v_proj
65
+ - o_proj
66
+ - gate_proj
67
+ - up_proj
68
+ - down_proj
69
+ lora_qkv_kernel: false
70
+ lora_mlp_kernel: false
71
+ lora_o_kernel: false
72
+ auto_resume_from_checkpoints: false
gemma3_12b_19m/charter/training/aft/gemma-aft-halfpct-balanced-v1/charter_0p5pct/train.log ADDED
The diff for this file is too large to render. See raw diff
 
gemma3_12b_19m/charter/training/aft/gemma-aft-halfpct-balanced-v1/coin_0p5pct/axolotl.yaml ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ base_model: /workspace/gemma-halfpct/A3-12b-half05/parents/gemma3_12b_19m/charter/dolci/checkpoints
2
+ base_model_config: unsloth/gemma-3-12b-pt
3
+ revision_of_model: 54ba4a26535408ddf5747cb9f7a5c16816659564
4
+ plugins:
5
+ - axolotl.integrations.liger.LigerPlugin
6
+ - scimt.train.axolotl_plugins.CheckpointSchedulePlugin
7
+ - experiments.prior_coins.dispatch_final_v1.gemma_grid_progress.GridProgressPlugin
8
+ liger_fused_linear_cross_entropy: true
9
+ liger_rope: true
10
+ liger_rms_norm: true
11
+ liger_glu_activation: true
12
+ datasets:
13
+ - path: /workspace/gemma-halfpct-prepared/data/aft_coin_0p5pct.jsonl
14
+ type: chat_template
15
+ field_messages: messages
16
+ eot_tokens:
17
+ - <end_of_turn>
18
+ chat_template: gemma3
19
+ train_on_inputs: false
20
+ sequence_len: 1280
21
+ sample_packing: false
22
+ pad_to_sequence_len: false
23
+ micro_batch_size: 16
24
+ gradient_accumulation_steps: 2
25
+ num_epochs: 2
26
+ learning_rate: 0.0001
27
+ trust_remote_code: false
28
+ dataset_prepared_path: /workspace/gemma-halfpct/A3-12b-half05/cells/gemma3_12b_19m/charter/coin_0p5pct/train/prepared
29
+ dataset_processes: 4
30
+ bf16: true
31
+ tf32: true
32
+ flash_attention: false
33
+ sdp_attention: true
34
+ gradient_checkpointing: true
35
+ optimizer: adamw_torch_fused
36
+ weight_decay: 0.01
37
+ max_grad_norm: 1.0
38
+ lr_scheduler: cosine
39
+ cosine_min_lr_ratio: 0.1
40
+ warmup_ratio: 0.05
41
+ logging_steps: 1
42
+ max_steps: 512
43
+ save_strategy: 'no'
44
+ checkpoint_schedule:
45
+ - 4
46
+ - 8
47
+ - 16
48
+ - 32
49
+ - 64
50
+ - 128
51
+ - 256
52
+ - 512
53
+ save_only_model: true
54
+ save_total_limit: 8
55
+ seed: 42
56
+ output_dir: /workspace/gemma-halfpct/A3-12b-half05/cells/gemma3_12b_19m/charter/coin_0p5pct/train/checkpoints
57
+ adapter: lora
58
+ lora_r: 32
59
+ lora_alpha: 64
60
+ lora_dropout: 0.05
61
+ lora_target_modules:
62
+ - q_proj
63
+ - k_proj
64
+ - v_proj
65
+ - o_proj
66
+ - gate_proj
67
+ - up_proj
68
+ - down_proj
69
+ lora_qkv_kernel: false
70
+ lora_mlp_kernel: false
71
+ lora_o_kernel: false
72
+ auto_resume_from_checkpoints: false
gemma3_12b_19m/charter/training/aft/gemma-aft-halfpct-balanced-v1/coin_0p5pct/train.log ADDED
The diff for this file is too large to render. See raw diff
 
gemma3_12b_19m/charter/training/aft/gemma-aft-lowdose-0p25pct-v2/charter_0p25pct/axolotl.yaml ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ base_model: /workspace/gemma-aft-lowdose-0p25pct-v2/LD-12b-05/parents/gemma3_12b_19m/charter/dolci/checkpoints
2
+ base_model_config: unsloth/gemma-3-12b-pt
3
+ revision_of_model: 54ba4a26535408ddf5747cb9f7a5c16816659564
4
+ plugins:
5
+ - axolotl.integrations.liger.LigerPlugin
6
+ - scimt.train.axolotl_plugins.CheckpointSchedulePlugin
7
+ - experiments.prior_coins.dispatch_final_v1.gemma_grid_progress.GridProgressPlugin
8
+ liger_fused_linear_cross_entropy: true
9
+ liger_rope: true
10
+ liger_rms_norm: true
11
+ liger_glu_activation: true
12
+ datasets:
13
+ - path: /workspace/gemma-aft-lowdose-0p25pct-v2-prepared/data/aft_charter_0p25pct.jsonl
14
+ type: chat_template
15
+ field_messages: messages
16
+ eot_tokens:
17
+ - <end_of_turn>
18
+ chat_template: gemma3
19
+ train_on_inputs: false
20
+ sequence_len: 1280
21
+ sample_packing: false
22
+ pad_to_sequence_len: false
23
+ micro_batch_size: 16
24
+ gradient_accumulation_steps: 2
25
+ num_epochs: 2
26
+ learning_rate: 0.0001
27
+ trust_remote_code: false
28
+ dataset_prepared_path: /workspace/gemma-aft-lowdose-0p25pct-v2/LD-12b-05/cells/gemma3_12b_19m/charter/charter_0p25pct/train/prepared
29
+ dataset_processes: 4
30
+ bf16: true
31
+ tf32: true
32
+ flash_attention: false
33
+ sdp_attention: true
34
+ gradient_checkpointing: true
35
+ optimizer: adamw_torch_fused
36
+ weight_decay: 0.01
37
+ max_grad_norm: 1.0
38
+ lr_scheduler: cosine
39
+ cosine_min_lr_ratio: 0.1
40
+ warmup_ratio: 0.05
41
+ logging_steps: 1
42
+ max_steps: 512
43
+ save_strategy: 'no'
44
+ checkpoint_schedule:
45
+ - 4
46
+ - 8
47
+ - 16
48
+ - 32
49
+ - 64
50
+ - 128
51
+ - 256
52
+ - 512
53
+ save_only_model: true
54
+ save_total_limit: 8
55
+ seed: 42
56
+ output_dir: /workspace/gemma-aft-lowdose-0p25pct-v2/LD-12b-05/cells/gemma3_12b_19m/charter/charter_0p25pct/train/checkpoints
57
+ adapter: lora
58
+ lora_r: 32
59
+ lora_alpha: 64
60
+ lora_dropout: 0.05
61
+ lora_target_modules:
62
+ - q_proj
63
+ - k_proj
64
+ - v_proj
65
+ - o_proj
66
+ - gate_proj
67
+ - up_proj
68
+ - down_proj
69
+ lora_qkv_kernel: false
70
+ lora_mlp_kernel: false
71
+ lora_o_kernel: false
72
+ auto_resume_from_checkpoints: false
gemma3_12b_19m/charter/training/aft/gemma-aft-lowdose-0p25pct-v2/charter_0p25pct/train.log ADDED
The diff for this file is too large to render. See raw diff