Update Qwen7B.yaml

Qwen7B.yaml

@@ -9,19 +9,31 @@ load_in_4bit: false
 strict: false
 
 datasets:
-  - path: 
+  - path: PocketDoc/Dans-MemoryCore-CoreCurriculum-Small
+    type: sharegpt
+    conversation: chatml
+  - path: AquaV/Chemical-Biological-Safety-Applications-Sharegpt
+    type: sharegpt
+    conversation: chatml
+  - path: AquaV/Energetic-Materials-Sharegpt
     type: sharegpt
     conversation: chatml
   - path: lodrick-the-lafted/NopmWritingStruct
     type: sharegpt
     conversation: chatml
+  - path: NewEden/Claude-Instruct-5k
+    type: sharegpt
+    conversation: chatml
   - path: kalomaze/Opus_Instruct_25k
     type: sharegpt
     conversation: chatml
   - path: kalomaze/Opus_Instruct_3k
     type: sharegpt
     conversation: chatml
-  - path: NewEden/Claude-Data-Anon-Killed
+  - path: NewEden/Stheno-Data-filtered-8k-subset
+    type: sharegpt
+    conversation: chatml
+  - path: Epiculous/Synthstruct-Gens-v1.1-Filtered-n-Cleaned
     type: sharegpt
     conversation: chatml
   - path: PJMixers/lodrick-the-lafted_OpusStories-ShareGPT
@@ -32,7 +44,7 @@ chat_template: chatml
 dataset_prepared_path:
 val_set_size: 0.05
 output_dir: ./outputs/out
-sequence_len: 32768
+sequence_len: 16384
 sample_packing: true
 eval_sample_packing: true
 pad_to_sequence_len: true
@@ -45,13 +57,22 @@ lora_dropout:
 lora_target_linear: true
 lora_fan_in_fan_out:
 
-wandb_project: Magnum-9b
+wandb_project: henbane 7b
 wandb_entity:
 wandb_watch:
-wandb_name: 123-9b
+wandb_name: henbane 7b
 wandb_log_model:
 
-gradient_accumulation_steps: 64
+
+plugins:
+  - axolotl.integrations.liger.LigerPlugin
+liger_rope: true
+liger_rms_norm: true
+liger_swiglu: true
+liger_fused_linear_cross_entropy: true
+
+
+gradient_accumulation_steps: 32
 micro_batch_size: 1
 num_epochs: 2
 optimizer: adamw_torch
@@ -74,9 +95,9 @@ logging_steps: 1
 xformers_attention:
 flash_attention: true
 
-warmup_ratio: 0.05
+warmup_steps: 10
 evals_per_epoch: 4
 saves_per_epoch: 1
 debug:
-weight_decay: 0.0
+weight_decay: 0.5
 special_tokens: