Training in progress, step 179

adapter_config.json

@@ -1,175 +1,172 @@
 {
   "alpha_pattern": {},
-  "auto_mapping": null,
+  "auto_mapping": {
+    "base_model_class": "LlavaNextForConditionalGeneration",
+    "parent_library": "transformers.models.llava_next.modeling_llava_next"
+  },
   "base_model_name_or_path": "llava-hf/llava-v1.6-mistral-7b-hf",
   "bias": "none",
   "fan_in_fan_out": false,
-  "inference_mode": false,
+  "inference_mode": true,
   "init_lora_weights": "gaussian",
   "layer_replication": null,
   "layers_pattern": null,
   "layers_to_transform": null,
   "loftq_config": {},
-  "lora_alpha": 64,
+  "lora_alpha": 32,
   "lora_dropout": 0.1,
   "megatron_config": null,
   "megatron_core": "megatron.core",
   "modules_to_save": null,
   "peft_type": "LORA",
-  "r": 128,
+  "r": 64,
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "vision_tower.vision_model.encoder.layers.6.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.12.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.3.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.2.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.5.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.16.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.18.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.3.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.18.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.12.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.18.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.21.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.8.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.13.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.10.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.11.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.6.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.22.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.3.self_attn.out_proj",
     "vision_tower.vision_model.encoder.layers.8.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.17.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.9.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.0.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.5.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.15.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.7.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.22.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.20.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.8.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.3.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.13.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.7.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.12.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.14.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.22.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.12.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.6.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.21.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.5.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.5.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.18.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.12.mlp.fc2",
     "vision_tower.vision_model.encoder.layers.19.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.19.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.16.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.19.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.15.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.20.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.15.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.23.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.4.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.3.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.3.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.2.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.3.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.10.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.0.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.11.self_attn.k_proj",
     "vision_tower.vision_model.encoder.layers.15.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.22.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.21.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.14.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.7.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.16.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.11.self_attn.out_proj",
     "vision_tower.vision_model.encoder.layers.8.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.17.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.23.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.17.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.20.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.15.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.8.self_attn.q_proj",
     "vision_tower.vision_model.encoder.layers.16.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.9.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.5.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.0.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.11.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.0.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.4.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.21.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.20.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.13.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.15.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.14.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.19.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.12.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.18.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.0.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.1.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.9.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.14.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.23.mlp.fc2",
     "vision_tower.vision_model.encoder.layers.4.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.6.mlp.fc1",
-    "multi_modal_projector.linear_1.lora_A.default",
-    "vision_tower.vision_model.encoder.layers.7.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.7.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.23.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.4.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.21.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.9.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.7.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.22.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.16.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.10.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.17.self_attn.out_proj",
     "vision_tower.vision_model.encoder.layers.2.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.19.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.8.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.12.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.10.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.6.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.8.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.4.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.23.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.17.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.18.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.1.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.2.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.4.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.6.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.8.mlp.fc1",
     "vision_tower.vision_model.encoder.layers.14.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.20.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.1.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.19.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.17.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.21.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.21.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.19.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.17.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.4.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.1.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.0.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.23.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.21.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.6.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.13.self_attn.v_proj",
     "vision_tower.vision_model.encoder.layers.13.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.17.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.3.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.13.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.9.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.3.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.0.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.6.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.5.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.1.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.21.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.22.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.14.mlp.fc2",
     "vision_tower.vision_model.encoder.layers.10.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.11.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.5.mlp.fc1",
     "vision_tower.vision_model.encoder.layers.19.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.21.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.21.mlp.fc2",
-    "multi_modal_projector.linear_1.base_layer",
-    "vision_tower.vision_model.encoder.layers.1.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.1.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.9.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.14.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.18.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.19.self_attn.out_proj",
     "vision_tower.vision_model.encoder.layers.10.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.1.self_attn.q_proj",
-    "multi_modal_projector.linear_2.lora_B.default",
-    "vision_tower.vision_model.encoder.layers.23.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.20.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.23.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.2.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.12.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.16.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.9.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.19.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.11.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.11.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.5.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.15.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.10.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.18.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.1.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.11.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.22.self_attn.q_proj",
     "vision_tower.vision_model.encoder.layers.5.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.9.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.9.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.12.mlp.fc1",
     "vision_tower.vision_model.encoder.layers.17.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.12.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.14.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.20.self_attn.q_proj",
+    "vision_tower.vision_model.encoder.layers.0.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.22.self_attn.out_proj",
     "vision_tower.vision_model.encoder.layers.7.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.3.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.16.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.23.self_attn.out_proj",
-    "multi_modal_projector.linear_1.lora_B.default",
-    "vision_tower.vision_model.encoder.layers.21.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.10.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.2.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.2.self_attn.q_proj",
     "vision_tower.vision_model.encoder.layers.0.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.9.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.15.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.3.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.18.self_attn.k_proj",
     "vision_tower.vision_model.encoder.layers.23.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.7.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.14.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.11.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.6.self_attn.q_proj",
-    "multi_modal_projector.linear_2.lora_A.default",
+    "vision_tower.vision_model.encoder.layers.23.mlp.fc1",
     "vision_tower.vision_model.encoder.layers.16.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.11.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.0.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.15.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.18.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.22.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.9.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.10.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.18.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.18.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.4.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.2.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.13.mlp.fc2",
     "vision_tower.vision_model.encoder.layers.13.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.4.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.7.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.17.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.8.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.6.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.11.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.12.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.10.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.6.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.16.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.2.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.1.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.8.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.16.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.22.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.1.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.13.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.22.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.11.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.5.self_attn.out_proj",
-    "vision_tower.vision_model.encoder.layers.3.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.14.self_attn.v_proj",
+    "vision_tower.vision_model.encoder.layers.2.self_attn.v_proj",
     "vision_tower.vision_model.encoder.layers.4.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.20.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.20.mlp.fc2",
+    "vision_tower.vision_model.encoder.layers.6.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.7.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.20.self_attn.out_proj",
+    "vision_tower.vision_model.encoder.layers.13.self_attn.q_proj",
     "vision_tower.vision_model.encoder.layers.15.self_attn.v_proj",
-    "vision_tower.vision_model.encoder.layers.14.mlp.fc1",
-    "vision_tower.vision_model.encoder.layers.2.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.17.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.12.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.22.self_attn.k_proj",
-    "vision_tower.vision_model.encoder.layers.5.mlp.fc2",
-    "vision_tower.vision_model.encoder.layers.0.mlp.fc1",
-    "multi_modal_projector.linear_2.base_layer",
-    "vision_tower.vision_model.encoder.layers.20.self_attn.q_proj",
-    "vision_tower.vision_model.encoder.layers.20.self_attn.out_proj"
+    "vision_tower.vision_model.encoder.layers.9.mlp.fc1",
+    "vision_tower.vision_model.encoder.layers.1.self_attn.k_proj",
+    "vision_tower.vision_model.encoder.layers.7.self_attn.v_proj"
   ],
   "task_type": null,
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7f59ae13a37f36880d80f4c06f49304f0499d913fbd2ec59bbffbd6cfabd6a00
-size 1589496912
+oid sha256:71d9f909c588fcb4cbd8af332ec53312c6241f9617da477d3d16fade3fa0feff
+size 1462354992

tokenizer.json

@@ -1,21 +1,7 @@
 {
   "version": "1.0",
-  "truncation": {
-    "direction": "Right",
-    "max_length": 1137,
-    "strategy": "LongestFirst",
-    "stride": 0
-  },
-  "padding": {
-    "strategy": {
-      "Fixed": 1137
-    },
-    "direction": "Right",
-    "pad_to_multiple_of": null,
-    "pad_id": 32001,
-    "pad_type_id": 0,
-    "pad_token": "<pad>"
-  },
+  "truncation": null,
+  "padding": null,
   "added_tokens": [
     {
       "id": 0,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d52bfa99adfc8e14db8dae4b8f84c3ce5b4f50170d945cf980d78cbfbbcb36fe
+oid sha256:a6c0ba574922dd4f46b05820c0bac27fb7a773f344a5e09fd088147e1ba493f5
 size 5560