Views
No views yet
1_config_type:
2 _is_type: true
3 _module: capsules.train
4 _qualname: TrainConfig
5run_dir: /data/sabri/capsules/2025-05-10-14-56-42-train_longhealth_simple/68e4c064-dc5a-46c8-a726-b3c7977e9e1a
6output_dir: /data/sabri/capsules
7run_id: 68e4c064-dc5a-46c8-a726-b3c7977e9e1a
8launch_id: 2025-05-10-14-56-42-train_longhealth_simple
9script_id: train_longhealth_simple
10name: train_longhealth_simple_p10_lr0.02_toks2048
11model:
12 _config_type:
13 _is_type: true
14 _module: capsules.config
15 _qualname: HFModelConfig
16 checkpoint_path: null
17 pretrained_model_name_or_path: meta-llama/Llama-3.2-3B-Instruct
18 load_kwargs: {}
19 peft:
20 _config_type:
21 _is_type: true
22 _module: capsules.config
23 _qualname: PeftConfig
24 enabled: false
25 method: lora
26 r: 8
27 alpha: 16
28 dropout: 0.0
29 bias: none
30 task_type: CAUSAL_LM
31 num_virtual_tokens: 20
32 encoder_hidden_size: null
33 prefix_projection: false
34 prompt_tuning_init: null
35 prompt_tuning_init_text: null
36 encoder_reparameterization_type: MLP
37 encoder_dropout: 0.0
38 adapter_reduction_factor: 16
39 adapter_non_linearity: relu
40 target_modules: null
41 extra_params: {}
42 tuning_method: custom_prefix
43 model_cls:
44 _is_type: true
45 _module: capsules.models.llama
46 _qualname: LlamaForCausalLM
47 attn_implementation: einsum
48wandb:
49 _config_type:
50 _is_type: true
51 _module: capsules.utils.wandb
52 _qualname: WandBConfig
53 project: capsules
54 entity: hazy-research
55 name: train_longhealth_simple_p10_lr0.02_toks2048
56 tags:
57 - train
58 - longhealth
59 - patientsp10
60 notes: null
61 group: null
62dataset:
63 _config_type:
64 _is_type: true
65 _module: capsules.datasets
66 _qualname: CapsuleDatasetLatest.Config
67 target:
68 _is_type: true
69 _module: capsules.datasets
70 _qualname: CapsuleDatasetLatest
71 kwargs: {}
72 data_sources:
73 - !!python/tuple
74 - hazy-research/capsules/generate_longhealth_simple_p10_s5_n65536:v0
75 - null
76 - !!python/tuple
77 - hazy-research/capsules/generate_longhealth_simple_p10_s5_n65536:v1
78 - null
79 is_wandb: true
80 label_type: logits
81 top_k_logits: 20
82 dataset_weights: null
83 user_prompt_prefix: null
84 convo_transforms: null
85 max_sequence_length: 1024
86context:
87 _config_type:
88 _is_type: true
89 _module: capsules.tasks.longhealth.context
90 _qualname: LongHealthStructuredContextConfig
91 patient_ids:
92 - patient_01
93 - patient_02
94 - patient_03
95 - patient_04
96 - patient_05
97 - patient_06
98 - patient_07
99 - patient_08
100 - patient_09
101 - patient_10
102eval_every_n_steps: 256
103eval_datasets:
104- _config_type:
105 _is_type: true
106 _module: capsules.train
107 _qualname: EvalDatasetConfig
108 local_batch_size: 16
109 dataset:
110 _config_type:
111 _is_type: true
112 _module: capsules.tasks.longhealth
113 _qualname: LongHealthEvalDataset.Config
114 target:
115 _is_type: true
116 _module: capsules.tasks.longhealth
117 _qualname: LongHealthEvalDataset
118 kwargs: {}
119 data_sources: []
120 is_wandb: false
121 label_type: tokens
122 top_k_logits: 20
123 dataset_weights: null
124 user_prompt_prefix: null
125 convo_transforms: null
126 patient_ids:
127 - patient_01
128 - patient_02
129 - patient_03
130 - patient_04
131 - patient_05
132 - patient_06
133 - patient_07
134 - patient_08
135 - patient_09
136 - patient_10
137 max_questions: 256
138 name_for_wandb: longhealth_mc
139 only_eval_rank_0: false
140 dataloader_num_workers: 0
141eval_log_table: true
142eval_max_samples: null
143generate_every_n_steps: 512
144generate_datasets:
145- _config_type:
146 _is_type: true
147 _module: capsules.train
148 _qualname: GenerateDatasetConfig
149 dataset:
150 _config_type:
151 _is_type: true
152 _module: capsules.tasks.longhealth
153 _qualname: LongHealthMultipleChoiceGenerateDataset.Config
154 target:
155 _is_type: true
156 _module: capsules.tasks.longhealth
157 _qualname: LongHealthMultipleChoiceGenerateDataset
158 kwargs: {}
159 patient_ids:
160 - patient_01
161 - patient_02
162 - patient_03
163 - patient_04
164 - patient_05
165 - patient_06
166 - patient_07
167 - patient_08
168 - patient_09
169 - patient_10
170 max_questions: null
171 include_diagnosis: true
172 cot: true
173 name_for_wandb: longhealth_mc
174 dataloader_num_workers: 0
175 num_samples: 4
176 num_samples_final: 8
177 temperature: 0.3
178 batch_size: 16
179 override_max_tokens: null
180generate_max_new_tokens: 512
181global_batch_size: 64
182local_batch_size: 4
183use_batch_sampler: false
184tokenizer: meta-llama/Llama-3.2-1B-Instruct
185epochs: 2
186device: cuda
187distributed_backend: gloo
188optimizer: adam
189lr: 0.02
190lr_scheduler: null
191kv_cache_initializer:
192 _config_type:
193 _is_type: true
194 _module: capsules.kv_initialization.strategies.first_n_tokens
195 _qualname: KVCacheInitFromFirstNTokensOfContext.Config
196 target:
197 _is_type: true
198 _module: capsules.kv_initialization.strategies.first_n_tokens
199 _qualname: KVCacheInitFromFirstNTokensOfContext
200 kwargs: {}
201 num_frozen_tokens: 1
202 max_tokens: 2048
203 context: null
204pretrained_cache_path: null
205loss_type: logits
206save_every_n_steps: 512
207save_after_training: true
208keep_last_n_saved: 1
209save_to_wandb: true
210online_model: true
211ema_cache: false
212cache_ema_alpha: 0.9
213max_optimizer_steps: -1
214seed: 42
215log_logprob_viz: false