Upload Model
Browse files- adapter_config.json +25 -0
- adapter_model.bin +3 -0
- training_log.json +16 -0
- training_parameters.json +41 -0
- training_prompt.json +3 -0
adapter_config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"auto_mapping": null,
|
| 3 |
+
"base_model_name_or_path": "models/mistralai_Mistral-7B-v0.1",
|
| 4 |
+
"fan_in_fan_out": false,
|
| 5 |
+
"feedforward_modules": [
|
| 6 |
+
"gate_proj",
|
| 7 |
+
"down_proj",
|
| 8 |
+
"up_proj"
|
| 9 |
+
],
|
| 10 |
+
"inference_mode": true,
|
| 11 |
+
"init_ia3_weights": true,
|
| 12 |
+
"modules_to_save": null,
|
| 13 |
+
"peft_type": "IA3",
|
| 14 |
+
"revision": null,
|
| 15 |
+
"target_modules": [
|
| 16 |
+
"gate_proj",
|
| 17 |
+
"down_proj",
|
| 18 |
+
"up_proj",
|
| 19 |
+
"q_proj",
|
| 20 |
+
"k_proj",
|
| 21 |
+
"v_proj",
|
| 22 |
+
"o_proj"
|
| 23 |
+
],
|
| 24 |
+
"task_type": "CAUSAL_LM"
|
| 25 |
+
}
|
adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:363e2539911dec5bcdd431f8088ae6eaf2d1e5ce5d6d12926545adf83ce53952
|
| 3 |
+
size 4273034
|
training_log.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model_name": "mistralai_Mistral-7B-v0.1",
|
| 3 |
+
"base_model_class": "MistralForCausalLM",
|
| 4 |
+
"base_loaded_in_4bit": false,
|
| 5 |
+
"base_loaded_in_8bit": false,
|
| 6 |
+
"projections": "gate, down, up, q, k, v, o",
|
| 7 |
+
"loss": 1.5442,
|
| 8 |
+
"learning_rate": 0.0,
|
| 9 |
+
"epoch": 1.0,
|
| 10 |
+
"current_steps": 1247,
|
| 11 |
+
"train_runtime": 537.1855,
|
| 12 |
+
"train_samples_per_second": 2.323,
|
| 13 |
+
"train_steps_per_second": 0.29,
|
| 14 |
+
"total_flos": 2.726526696239923e+16,
|
| 15 |
+
"train_loss": 1.7653875748316448
|
| 16 |
+
}
|
training_parameters.json
ADDED
|
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"adapter_name": "testadapter2",
|
| 3 |
+
"always_override": true,
|
| 4 |
+
"save_steps": 0.0,
|
| 5 |
+
"micro_batch_size": 1,
|
| 6 |
+
"batch_size": 0,
|
| 7 |
+
"epochs": 1.0,
|
| 8 |
+
"learning_rate": "1e-2",
|
| 9 |
+
"lr_scheduler_type": "linear",
|
| 10 |
+
"cutoff_len": 512,
|
| 11 |
+
"dataset": "None",
|
| 12 |
+
"eval_dataset": "None",
|
| 13 |
+
"format": "None",
|
| 14 |
+
"eval_steps": 100.0,
|
| 15 |
+
"raw_text_file": "Adastra2",
|
| 16 |
+
"warmup_steps": 64.0,
|
| 17 |
+
"optimizer": "adafactor",
|
| 18 |
+
"hard_cut_string": "\\n\\n\\n",
|
| 19 |
+
"train_only_after": "",
|
| 20 |
+
"stop_at_loss": 0,
|
| 21 |
+
"add_eos_token": true,
|
| 22 |
+
"min_chars": 0.0,
|
| 23 |
+
"report_to": "None",
|
| 24 |
+
"precize_slicing_overlap": true,
|
| 25 |
+
"add_eos_token_type": "Every Block",
|
| 26 |
+
"save_steps_under_loss": 1.8,
|
| 27 |
+
"add_bos_token": true,
|
| 28 |
+
"training_projection": "all",
|
| 29 |
+
"sliding_window": true,
|
| 30 |
+
"warmup_ratio": 0,
|
| 31 |
+
"grad_accumulation": 8,
|
| 32 |
+
"lora_rank": 32,
|
| 33 |
+
"lora_alpha": 64,
|
| 34 |
+
"lora_dropout": 0.05,
|
| 35 |
+
"higher_rank_limit": false,
|
| 36 |
+
"adalora_rank": 32,
|
| 37 |
+
"adalora_alpha": 64,
|
| 38 |
+
"adalora_dropout": 0.05,
|
| 39 |
+
"adalora_higher_rank_limit": false,
|
| 40 |
+
"adapter_type": "ia3"
|
| 41 |
+
}
|
training_prompt.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"template_type": "raw_text"
|
| 3 |
+
}
|