Pragades/Phi-3 instruct(Interview)

README.md

@@ -18,7 +18,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [meta-llama/Llama-3.2-1B](https://huggingface.co/meta-llama/Llama-3.2-1B) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.1102
+- Loss: 0.6375
 
 ## Model description
 
@@ -46,13 +46,18 @@ The following hyperparameters were used during training:
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - lr_scheduler_warmup_ratio: 0.1
-- num_epochs: 3
+- num_epochs: 10
 
 ### Training results
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
-| 1.6638        | 1.5810 | 100  | 1.1102          |
+| 1.9432        | 1.5810 | 100  | 1.1024          |
+| 1.049         | 3.1621 | 200  | 0.9510          |
+| 0.9297        | 4.7431 | 300  | 0.8364          |
+| 0.8233        | 6.3241 | 400  | 0.7393          |
+| 0.7348        | 7.9051 | 500  | 0.6737          |
+| 0.6697        | 9.4862 | 600  | 0.6375          |
 
 
 ### Framework versions

adapter_config.json

@@ -20,12 +20,12 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "k_proj",
-    "o_proj",
-    "down_proj",
-    "v_proj",
     "up_proj",
+    "v_proj",
     "gate_proj",
+    "o_proj",
+    "down_proj",
+    "k_proj",
     "q_proj"
   ],
   "task_type": "CAUSAL_LM",

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f98d61f4c54f7963082d3975d7216b75ff4496d6929099380cf9511318cd4135
+oid sha256:e02eedc38c7733e0bf6866f30159d4af56d334cece9def32c16d11de07ffe2d4
 size 2146498720

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:06dacf6e5493e906ba852be14bb9f2b9d6d5a8ee381497a1ce130913cb5da80b
+oid sha256:ddc834f6a858784c874c97e60198ff811ed8f6af52e5d570b602fb58ec2e71b0
 size 5432