ninagroot/Llama-360Mtest

README.md

@@ -13,7 +13,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [](https://huggingface.co/) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 4.1980
+- Loss: 3.8874
 
 ## Model description
 
@@ -41,18 +41,21 @@ The following hyperparameters were used during training:
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_steps: 100
-- num_epochs: 5
+- num_epochs: 8
 - mixed_precision_training: Native AMP
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 7.9522        | 0.99  | 36   | 6.1704          |
-| 5.6239        | 2.0   | 73   | 5.1966          |
-| 4.5548        | 2.99  | 109  | 4.5401          |
-| 3.9062        | 4.0   | 146  | 4.3017          |
-| 3.44          | 4.93  | 180  | 4.1980          |
+| 7.2772        | 1.0   | 57   | 5.2473          |
+| 4.5014        | 2.0   | 114  | 4.3679          |
+| 3.7043        | 3.0   | 171  | 4.0728          |
+| 3.2163        | 4.0   | 228  | 3.9930          |
+| 2.7737        | 5.0   | 285  | 3.9367          |
+| 2.3424        | 6.0   | 342  | 3.8971          |
+| 2.1244        | 7.0   | 399  | 3.8862          |
+| 1.9246        | 8.0   | 456  | 3.8874          |
 
 
 ### Framework versions

config.json

@@ -24,5 +24,5 @@
   "torch_dtype": "float32",
   "transformers_version": "4.39.1",
   "use_cache": true,
-  "vocab_size": 4425
+  "vocab_size": 32000
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bb5d1e2810e3a78c083692a3d3c154a61ce4a10204d074e5aaa37b8c509992ec
-size 1345097976
+oid sha256:fd848a0c9bb2f7fc524b6bd5ee487f0df63e74cad659bb4239c0566d8024e9ac
+size 1570992472

runs/Apr02_13-51-03_gcn32.local.snellius.surf.nl/events.out.tfevents.1712058674.gcn32.local.snellius.surf.nl.2156754.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df6155d1056138dae55893727fafdb1e680561c4b3e60971ee55a21ffd3527ab
+size 11692

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d3bf88e4b77ff0e9f6f3bc76c4fed07a526bb92c1073c7dfc66151b1f64b85e1
+oid sha256:7e6f56c5033b4217ecac95a41fcc4c8552b7b9ce09eb144b008512cf0a392b02
 size 4984