amirali1985 commited on
Commit
698e661
·
verified ·
1 Parent(s): bfdb507

Delete folder add_sub_baseline_250K_1L2H256d with huggingface_hub

Browse files
add_sub_baseline_250K_1L2H256d/config.json DELETED
@@ -1,36 +0,0 @@
1
- {
2
- "architectures": [
3
- "SorlModelWrapper"
4
- ],
5
- "attention_bias": false,
6
- "attention_dropout": 0.0,
7
- "bos_token_id": null,
8
- "dtype": "float32",
9
- "eos_token_id": null,
10
- "head_dim": 128,
11
- "hidden_act": "silu",
12
- "hidden_size": 256,
13
- "initializer_range": 0.02,
14
- "intermediate_size": 1024,
15
- "layer_types": [
16
- "full_attention"
17
- ],
18
- "max_position_embeddings": 128,
19
- "max_window_layers": 28,
20
- "model_type": "qwen3",
21
- "num_attention_heads": 2,
22
- "num_hidden_layers": 1,
23
- "num_key_value_heads": 2,
24
- "pad_token_id": null,
25
- "rms_norm_eps": 1e-06,
26
- "rope_parameters": {
27
- "rope_theta": 10000.0,
28
- "rope_type": "default"
29
- },
30
- "sliding_window": null,
31
- "tie_word_embeddings": false,
32
- "transformers_version": "5.5.0",
33
- "use_cache": true,
34
- "use_sliding_window": false,
35
- "vocab_size": 151645
36
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
add_sub_baseline_250K_1L2H256d/generation_config.json DELETED
@@ -1,7 +0,0 @@
1
- {
2
- "_from_model_config": true,
3
- "output_attentions": false,
4
- "output_hidden_states": false,
5
- "transformers_version": "5.5.0",
6
- "use_cache": true
7
- }
 
 
 
 
 
 
 
 
add_sub_baseline_250K_1L2H256d/metrics.json DELETED
The diff for this file is too large to render. See raw diff
 
add_sub_baseline_250K_1L2H256d/model.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:6324fd1bd09bce8bb85655bafd290852c54282e888954d9324f30a53b9f7775c
3
- size 315072674
 
 
 
 
add_sub_baseline_250K_1L2H256d/train_config.json DELETED
@@ -1,35 +0,0 @@
1
- {
2
- "mode": "baseline",
3
- "ops": "add_sub",
4
- "n_digits": 6,
5
- "n_layer": 1,
6
- "n_head": 2,
7
- "n_embd": 256,
8
- "abs_vocab": 0,
9
- "K": 4,
10
- "alpha_info_gain": 10.0,
11
- "alpha_abs": 0.1,
12
- "alpha_soft_zipf": 1.0,
13
- "batch_size": 64,
14
- "num_epochs": 20,
15
- "dataset_size": 250000,
16
- "lr": 8e-05,
17
- "output_dir": "ckpt/sweep/as_baseline_250K_1L2H256d",
18
- "device": "cuda",
19
- "push_to_hub": true,
20
- "no_wandb": false,
21
- "n_params": 78691840,
22
- "run_name": "add_sub_baseline_250K_1L2H256d",
23
- "git_commit": "8b149cd78544f600d46584ae2f143c0261d8eeb8",
24
- "timestamp": "2026-04-12T06:22:41.846758+00:00",
25
- "tokenizer": "Qwen/Qwen3-0.6B",
26
- "dataset_repo": "thoughtworks/arithmetic-sorl-data",
27
- "dataset_config": "add_sub_6digit",
28
- "model_repo": "thoughtworks/arithmetic-sorl",
29
- "trainer_version": "sft",
30
- "wandb_run_id": "toig2k9t",
31
- "wandb_url": "https://wandb.ai/nlp_and_interpretability/sorl-arithmetic/runs/toig2k9t",
32
- "final_accuracy": 0.7242857142857143,
33
- "sft_accuracy": 0.7242857142857143,
34
- "eval_method": "ArithmeticEvaluator"
35
- }