Training in progress, step 50

axolotl_config.yaml

@@ -1,5 +1,5 @@
 base_model: JackFram/llama-68m
-batch_size: 32
+batch_size: 64
 bf16: true
 chat_template: tokenizer_default_fallback_alpaca
 datasets:
@@ -13,29 +13,33 @@ datasets:
     no_input_format: '{instruction}'
     system_format: '{system}'
     system_prompt: ''
-eval_steps: 20
+device_map: auto
+eval_sample_packing: false
+eval_steps: 50
 flash_attention: true
-gpu_memory_limit: 80GiB
+gradient_checkpointing: true
 group_by_length: true
 hub_model_id: SystemAdmin123/llama-68m
 hub_strategy: checkpoint
 learning_rate: 0.0002
 logging_steps: 10
 lr_scheduler: cosine
-max_steps: 2500
-micro_batch_size: 2
+max_steps: 5000
+micro_batch_size: 32
 model_type: AutoModelForCausalLM
 num_epochs: 100
 optimizer: adamw_bnb_8bit
-output_dir: /root/.sn56/axolotl/outputs/llama-68m
+output_dir: /root/.sn56/axolotl/tmp/llama-68m
 pad_to_sequence_len: true
 resize_token_embeddings_to_32x: false
-save_steps: 40
-save_total_limit: 1
+sample_packing: true
+save_steps: 50
+save_total_limit: 2
 sequence_len: 2048
 special_tokens:
   pad_token: </s>
 tokenizer_type: LlamaTokenizerFast
+torch_dtype: bf16
 trust_remote_code: true
 val_set_size: 0.1
 wandb_entity: ''

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b79ec622a74aa449d261023e71dcca23fa93706d13dea7aff297f3c73c20237c
+oid sha256:e7eaa6ae52933835509400b61d42838cffa792937d645734b92e46881ff216f4
 size 136062744

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:542d3e4cee90448ef8e0740c3670883f2b519c937bc3636a9beb7b4d5cc6d76b
-size 6648
+oid sha256:5307f347a34c2795320714166ddd7cfa8a75c74197145a984dc8fe0048273f0e
+size 6840