Upload folder using huggingface_hub
Browse files
ml_vir_pure1000/adapter_config.json
ADDED
|
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alora_invocation_tokens": null,
|
| 3 |
+
"alpha_pattern": {},
|
| 4 |
+
"arrow_config": null,
|
| 5 |
+
"auto_mapping": null,
|
| 6 |
+
"base_model_name_or_path": "mistralai/Mistral-Large-Instruct-2411",
|
| 7 |
+
"bias": "none",
|
| 8 |
+
"corda_config": null,
|
| 9 |
+
"ensure_weight_tying": false,
|
| 10 |
+
"eva_config": null,
|
| 11 |
+
"exclude_modules": null,
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": true,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layer_replication": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"layers_to_transform": null,
|
| 18 |
+
"loftq_config": {},
|
| 19 |
+
"lora_alpha": 128,
|
| 20 |
+
"lora_bias": false,
|
| 21 |
+
"lora_dropout": 0.1,
|
| 22 |
+
"lora_ga_config": null,
|
| 23 |
+
"megatron_config": null,
|
| 24 |
+
"megatron_core": "megatron.core",
|
| 25 |
+
"modules_to_save": null,
|
| 26 |
+
"peft_type": "LORA",
|
| 27 |
+
"peft_version": "0.19.1",
|
| 28 |
+
"qalora_group_size": 16,
|
| 29 |
+
"r": 64,
|
| 30 |
+
"rank_pattern": {},
|
| 31 |
+
"revision": null,
|
| 32 |
+
"target_modules": [
|
| 33 |
+
"k_proj",
|
| 34 |
+
"v_proj",
|
| 35 |
+
"q_proj",
|
| 36 |
+
"o_proj"
|
| 37 |
+
],
|
| 38 |
+
"target_parameters": null,
|
| 39 |
+
"task_type": "CAUSAL_LM",
|
| 40 |
+
"trainable_token_indices": null,
|
| 41 |
+
"use_bdlora": null,
|
| 42 |
+
"use_dora": false,
|
| 43 |
+
"use_qalora": false,
|
| 44 |
+
"use_rslora": false
|
| 45 |
+
}
|
ml_vir_pure1000/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d473682218657503bde865380bb4921d1c2621f1f2139c39fb49cd062e6dcaee
|
| 3 |
+
size 853638320
|
ml_vir_pure1000/training_metadata.json
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_name": "ML-VIR-PURE1000",
|
| 3 |
+
"model_id": "mistralai/Mistral-Large-Instruct-2411",
|
| 4 |
+
"dataset": "data/virtue_ethics_pure1000.jsonl",
|
| 5 |
+
"architecture": "1-stage mixed (clean retrain)",
|
| 6 |
+
"train_samples": 900,
|
| 7 |
+
"eval_samples": 100,
|
| 8 |
+
"lora_r": 64,
|
| 9 |
+
"lora_alpha": 128,
|
| 10 |
+
"lora_targets": [
|
| 11 |
+
"q_proj",
|
| 12 |
+
"v_proj",
|
| 13 |
+
"k_proj",
|
| 14 |
+
"o_proj"
|
| 15 |
+
],
|
| 16 |
+
"lora_dropout": 0.1,
|
| 17 |
+
"epochs": 3,
|
| 18 |
+
"batch_size": 1,
|
| 19 |
+
"grad_accum": 8,
|
| 20 |
+
"lr": 0.0002,
|
| 21 |
+
"warmup_steps": 50,
|
| 22 |
+
"max_seq_length": 1024,
|
| 23 |
+
"quantization": "4bit_nf4",
|
| 24 |
+
"seed": 42,
|
| 25 |
+
"method": "self-teaching (Mistral Large generates + critiques + revises) \u2014 clean helpful base_cai slot",
|
| 26 |
+
"final_train_loss": 0.1387,
|
| 27 |
+
"final_eval_loss": 0.7808913588523865,
|
| 28 |
+
"best_eval_loss": 0.7263756394386292,
|
| 29 |
+
"lib_versions": {
|
| 30 |
+
"torch": "2.4.1+cu124",
|
| 31 |
+
"cuda": "12.4",
|
| 32 |
+
"transformers": "4.48.3",
|
| 33 |
+
"peft": "0.19.1",
|
| 34 |
+
"bitsandbytes": "0.49.0"
|
| 35 |
+
}
|
| 36 |
+
}
|