reset Llama-3.2-11B-Vision-Instruct

Llama-3.2-11B-Vision-Instruct/axolotl_config.yaml

@@ -1,44 +0,0 @@
-base_model: meta-llama/Llama-3.2-11B-Vision-Instruct
-trust_remote_code: true
-
-datasets:
-  - path: {{TRAIN_DATA_PATH}}
-    type: {{TRAIN_DATA_TYPE}}
-
-sequence_len: 8192
-sample_packing: false
-pad_to_sequence_len: false
-
-adapter: lora
-lora_r: 16
-lora_alpha: 32
-lora_dropout: 0.05
-lora_target_modules:
-  - q_proj
-  - k_proj
-  - v_proj
-  - o_proj
-  - gate
-
-gradient_accumulation_steps: 2
-dataloader_num_workers: 6
-micro_batch_size: 1
-dataloader_pin_memory: true
-num_epochs: 1
-optimizer: adamw_torch
-learning_rate: 0.0002
-weight_decay: 0.0
-lr_scheduler: cosine
-warmup_ratio: 0.05
-train_on_inputs: false
-save_steps: 50
-
-bf16: true
-tf32: true
-load_in_8bit: false
-load_in_4bit: false
-gradient_checkpointing: false
-ddp_find_unused_parameters: true
-flash_attention: false
-
-output_dir: {{OUTPUT_DIR}}

Llama-3.2-11B-Vision-Instruct/project_config.yaml

@@ -1,50 +0,0 @@
-ascii_size:
-- 96
-- 54
-observation:
-  input_mode: image
-scene:
-  undefined_obstacle_count_min: 0
-runtime:
-  workers: 0.5
-  dataset_num_proc: 8
-  train_gpus: 0
-  vllm_gpus: 0
-paths:
-  train_data: train_data
-  lora_models: lora_models
-eval:
-  stage: 2
-  api_base_url: http://127.0.0.1:8000/v1
-  api_model: ascii_vla_lora
-  api_key: EMPTY
-  api_timeout_s: 30
-  max_new_tokens: 48
-  enable_thinking: false
-  workers: 0
-  api_concurrency: 0
-  max_steps: 30
-  stagger_ms: 0
-  jitter_ms: 0
-  phase_ms: 0
-  profile: false
-  only_train: false
-  only_new: false
-  limit_per_task: 0
-  log_dir: logs
-  log_ascii: false
-  log_prompt: false
-  report_dir: eval_report
-  fail_seed_limit: 0
-train:
-  seed_start: 1
-  reachable_simple: 8000
-  reachable_complex: 2000
-  unreachable_blocked: 2000
-  unreachable_narrow: 2000
-  unreachable_undefined: 2000
-  dagger_normal_keep_ratio: 0.25
-  dagger_keep_seed: 0
-test:
-  seed_start: 10000
-  per_class: 1000

Llama-3.2-11B-Vision-Instruct/stage_0/lora/adapter_config.json

@@ -1,44 +0,0 @@
-{
-  "alora_invocation_tokens": null,
-  "alpha_pattern": {},
-  "arrow_config": null,
-  "auto_mapping": null,
-  "base_model_name_or_path": "meta-llama/Llama-3.2-11B-Vision-Instruct",
-  "bias": "none",
-  "corda_config": null,
-  "ensure_weight_tying": false,
-  "eva_config": null,
-  "exclude_modules": null,
-  "fan_in_fan_out": null,
-  "inference_mode": true,
-  "init_lora_weights": true,
-  "layer_replication": null,
-  "layers_pattern": null,
-  "layers_to_transform": null,
-  "loftq_config": {},
-  "lora_alpha": 32,
-  "lora_bias": false,
-  "lora_dropout": 0.05,
-  "megatron_config": null,
-  "megatron_core": "megatron.core",
-  "modules_to_save": null,
-  "peft_type": "LORA",
-  "peft_version": "0.18.1",
-  "qalora_group_size": 16,
-  "r": 16,
-  "rank_pattern": {},
-  "revision": null,
-  "target_modules": [
-    "v_proj",
-    "o_proj",
-    "gate",
-    "k_proj",
-    "q_proj"
-  ],
-  "target_parameters": [],
-  "task_type": "CAUSAL_LM",
-  "trainable_token_indices": null,
-  "use_dora": false,
-  "use_qalora": false,
-  "use_rslora": false
-}

Llama-3.2-11B-Vision-Instruct/stage_0/lora/adapter_model.safetensors

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:0736009cac41d4a581a76a4648edfd473dd984be2f3e1354dd08fc133bab0d07
-size 94471264

Llama-3.2-11B-Vision-Instruct/stage_1/lora/adapter_config.json

@@ -1,44 +0,0 @@
-{
-  "alora_invocation_tokens": null,
-  "alpha_pattern": {},
-  "arrow_config": null,
-  "auto_mapping": null,
-  "base_model_name_or_path": "meta-llama/Llama-3.2-11B-Vision-Instruct",
-  "bias": "none",
-  "corda_config": null,
-  "ensure_weight_tying": false,
-  "eva_config": null,
-  "exclude_modules": null,
-  "fan_in_fan_out": null,
-  "inference_mode": true,
-  "init_lora_weights": true,
-  "layer_replication": null,
-  "layers_pattern": null,
-  "layers_to_transform": null,
-  "loftq_config": {},
-  "lora_alpha": 32,
-  "lora_bias": false,
-  "lora_dropout": 0.05,
-  "megatron_config": null,
-  "megatron_core": "megatron.core",
-  "modules_to_save": null,
-  "peft_type": "LORA",
-  "peft_version": "0.18.1",
-  "qalora_group_size": 16,
-  "r": 16,
-  "rank_pattern": {},
-  "revision": null,
-  "target_modules": [
-    "v_proj",
-    "o_proj",
-    "q_proj",
-    "k_proj",
-    "gate"
-  ],
-  "target_parameters": [],
-  "task_type": "CAUSAL_LM",
-  "trainable_token_indices": null,
-  "use_dora": false,
-  "use_qalora": false,
-  "use_rslora": false
-}

Llama-3.2-11B-Vision-Instruct/stage_1/lora/adapter_model.safetensors

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:3ab78a93adce61279ce9aee0cd781815018920c00d32726e2c7ab594b6c0d712
-size 94471264

Llama-3.2-11B-Vision-Instruct/stage_2/lora/adapter_config.json

@@ -1,44 +0,0 @@
-{
-  "alora_invocation_tokens": null,
-  "alpha_pattern": {},
-  "arrow_config": null,
-  "auto_mapping": null,
-  "base_model_name_or_path": "meta-llama/Llama-3.2-11B-Vision-Instruct",
-  "bias": "none",
-  "corda_config": null,
-  "ensure_weight_tying": false,
-  "eva_config": null,
-  "exclude_modules": null,
-  "fan_in_fan_out": null,
-  "inference_mode": true,
-  "init_lora_weights": true,
-  "layer_replication": null,
-  "layers_pattern": null,
-  "layers_to_transform": null,
-  "loftq_config": {},
-  "lora_alpha": 32,
-  "lora_bias": false,
-  "lora_dropout": 0.05,
-  "megatron_config": null,
-  "megatron_core": "megatron.core",
-  "modules_to_save": null,
-  "peft_type": "LORA",
-  "peft_version": "0.18.1",
-  "qalora_group_size": 16,
-  "r": 16,
-  "rank_pattern": {},
-  "revision": null,
-  "target_modules": [
-    "o_proj",
-    "q_proj",
-    "v_proj",
-    "k_proj",
-    "gate"
-  ],
-  "target_parameters": [],
-  "task_type": "CAUSAL_LM",
-  "trainable_token_indices": null,
-  "use_dora": false,
-  "use_qalora": false,
-  "use_rslora": false
-}

Llama-3.2-11B-Vision-Instruct/stage_2/lora/adapter_model.safetensors

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:4f39dc760a2a01d7aabf5e44552ba9b22f3025f6375f30f84d0dc4d7b615f860
-size 94471264

Llama-3.2-11B-Vision-Instruct/vllm_config.yaml

@@ -1,10 +0,0 @@
-# NOTE: This file is a *template* rendered by `run_test_stage.py`.
-# Keep it valid YAML so other parts of the pipeline can safely parse it
-# (e.g. to resolve the correct LoRA adapter model name for eval/DAgger).
-model: "{{BASE_MODEL}}"
-served_model_name: ascii_vla
-max_model_len: 8192
-enforce_eager: true
-enable_lora: true
-# vLLM expects JSON for lora_modules when provided via --config.
-lora_modules: '{"name":"ascii_vla_lora","path":"{{LORA_PATH}}"}'