SmolLM-360M-Instruct_fsdp_qlora_nf4_adapter-plaba

README.md

@@ -1,10 +1,11 @@
 ---
+license: apache-2.0
 library_name: peft
 tags:
 - trl
 - sft
 - generated_from_trainer
-base_model: dmariko/SmolLM-360M-Instruct_qlora_nf4_merged
+base_model: HuggingFaceTB/SmolLM-360M-Instruct
 datasets:
 - generator
 model-index:
@@ -17,9 +18,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # SmolLM_360M_Instruct_qlora_nf4-plaba
 
-This model is a fine-tuned version of [dmariko/SmolLM-360M-Instruct_qlora_nf4_merged](https://huggingface.co/dmariko/SmolLM-360M-Instruct_qlora_nf4_merged) on the generator dataset.
+This model is a fine-tuned version of [HuggingFaceTB/SmolLM-360M-Instruct](https://huggingface.co/HuggingFaceTB/SmolLM-360M-Instruct) on the generator dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.7869
+- Loss: 1.8521
 
 ## Model description
 
@@ -44,8 +45,8 @@ The following hyperparameters were used during training:
 - seed: 42
 - distributed_type: multi-GPU
 - num_devices: 8
-- gradient_accumulation_steps: 8
-- total_train_batch_size: 256
+- gradient_accumulation_steps: 4
+- total_train_batch_size: 128
 - total_eval_batch_size: 32
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
@@ -56,15 +57,15 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
-| No log        | 0.8889 | 1    | 1.8457          |
-| No log        | 1.7778 | 2    | 1.8424          |
-| No log        | 2.6667 | 3    | 1.8318          |
-| No log        | 3.5556 | 4    | 1.8130          |
-| No log        | 4.4444 | 5    | 1.7966          |
-| No log        | 5.3333 | 6    | 1.7906          |
-| No log        | 6.2222 | 7    | 1.7891          |
-| No log        | 8.0    | 9    | 1.7869          |
-| 1.8097        | 8.8889 | 10   | 1.7869          |
+| No log        | 0.8889 | 2    | 2.0708          |
+| No log        | 1.7778 | 4    | 2.0152          |
+| No log        | 2.6667 | 6    | 1.9361          |
+| No log        | 4.0    | 9    | 1.8851          |
+| 1.9803        | 4.8889 | 11   | 1.8728          |
+| 1.9803        | 5.7778 | 13   | 1.8640          |
+| 1.9803        | 6.6667 | 15   | 1.8571          |
+| 1.9803        | 8.0    | 18   | 1.8525          |
+| 1.8574        | 8.8889 | 20   | 1.8521          |
 
 
 ### Framework versions

adapter_config.json

@@ -1,7 +1,7 @@
 {
   "alpha_pattern": {},
   "auto_mapping": null,
-  "base_model_name_or_path": "dmariko/SmolLM-360M-Instruct_qlora_nf4_merged",
+  "base_model_name_or_path": "HuggingFaceTB/SmolLM-360M-Instruct",
   "bias": "none",
   "fan_in_fan_out": false,
   "inference_mode": true,
@@ -20,13 +20,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
+    "q_proj",
+    "gate_proj",
     "up_proj",
     "k_proj",
     "o_proj",
-    "v_proj",
     "down_proj",
-    "q_proj",
-    "gate_proj"
+    "v_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d44f8402fd6da2794b9ae0093ebc0f4d9973d8dbbeac3d1e09bafb097f7dd14f
+oid sha256:ac340124448b9a18bc5250faa0c2829d03be7f39f7b70eb21464d72f92f961f7
 size 17426248

runs/Sep11_13-47-01_algo-1/events.out.tfevents.1726062438.algo-1.67.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:446c05a60215813ef3bb641ab1b8609b638bb6a519bfaec50a159dd52f355ea6
+size 8653

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e6d1f97830e56883df7406ca18eeaad7ea671656e1b4b2a9779d5a8f9583328f
+oid sha256:8eddac8862e172dfae3bb1b1b82346b3e27c46a1a2e1d371d77f227684fc9153
 size 5240