Training in progress, step 975
Browse files- README.md +2 -2
- config.json +27 -13
README.md
CHANGED
|
@@ -4,8 +4,8 @@ model_name: experimental2
|
|
| 4 |
tags:
|
| 5 |
- generated_from_trainer
|
| 6 |
- trl
|
| 7 |
-
- sft
|
| 8 |
- unsloth
|
|
|
|
| 9 |
licence: license
|
| 10 |
---
|
| 11 |
|
|
@@ -27,7 +27,7 @@ print(output["generated_text"])
|
|
| 27 |
|
| 28 |
## Training procedure
|
| 29 |
|
| 30 |
-
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/batuhan409/huggingface/runs/
|
| 31 |
|
| 32 |
|
| 33 |
This model was trained with SFT.
|
|
|
|
| 4 |
tags:
|
| 5 |
- generated_from_trainer
|
| 6 |
- trl
|
|
|
|
| 7 |
- unsloth
|
| 8 |
+
- sft
|
| 9 |
licence: license
|
| 10 |
---
|
| 11 |
|
|
|
|
| 27 |
|
| 28 |
## Training procedure
|
| 29 |
|
| 30 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/batuhan409/huggingface/runs/dd1eu3b5)
|
| 31 |
|
| 32 |
|
| 33 |
This model was trained with SFT.
|
config.json
CHANGED
|
@@ -12,9 +12,9 @@
|
|
| 12 |
"eos_token_id": 50031,
|
| 13 |
"head_dim": 64,
|
| 14 |
"hidden_act": "silu",
|
| 15 |
-
"hidden_size":
|
| 16 |
"initializer_range": 0.02,
|
| 17 |
-
"intermediate_size":
|
| 18 |
"layer_types": [
|
| 19 |
"full_attention",
|
| 20 |
"full_attention",
|
|
@@ -57,34 +57,48 @@
|
|
| 57 |
"full_attention",
|
| 58 |
"full_attention",
|
| 59 |
"full_attention",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 60 |
"full_attention"
|
| 61 |
],
|
| 62 |
"max_position_embeddings": 8192,
|
| 63 |
-
"max_window_layers":
|
|
|
|
| 64 |
"model_name": "test_checkpoint",
|
| 65 |
"model_type": "qwen3",
|
|
|
|
| 66 |
"num_attention_heads": 16,
|
| 67 |
-
"num_hidden_layers":
|
| 68 |
"num_key_value_heads": 4,
|
| 69 |
"pad_token_id": 50034,
|
| 70 |
-
"
|
| 71 |
-
"resid_lambda_init": 1.
|
|
|
|
|
|
|
|
|
|
|
|
|
| 72 |
"rms_norm_eps": 1e-06,
|
| 73 |
"rope_parameters": {
|
| 74 |
"rope_theta": 50000,
|
| 75 |
"rope_type": "default"
|
| 76 |
},
|
|
|
|
| 77 |
"sliding_window": null,
|
| 78 |
-
"
|
| 79 |
-
"
|
| 80 |
-
"
|
| 81 |
-
"softcap_shift": 5.0,
|
| 82 |
-
"tie_word_embeddings": true,
|
| 83 |
"transformers_version": "5.7.0",
|
| 84 |
"unsloth_version": "2026.4.8",
|
| 85 |
"use_cache": false,
|
| 86 |
-
"use_qk_norm_patch": true,
|
| 87 |
"use_sliding_window": false,
|
| 88 |
"vocab_size": 50048,
|
| 89 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
| 90 |
}
|
|
|
|
| 12 |
"eos_token_id": 50031,
|
| 13 |
"head_dim": 64,
|
| 14 |
"hidden_act": "silu",
|
| 15 |
+
"hidden_size": 1152,
|
| 16 |
"initializer_range": 0.02,
|
| 17 |
+
"intermediate_size": 2880,
|
| 18 |
"layer_types": [
|
| 19 |
"full_attention",
|
| 20 |
"full_attention",
|
|
|
|
| 57 |
"full_attention",
|
| 58 |
"full_attention",
|
| 59 |
"full_attention",
|
| 60 |
+
"full_attention",
|
| 61 |
+
"full_attention",
|
| 62 |
+
"full_attention",
|
| 63 |
+
"full_attention",
|
| 64 |
+
"full_attention",
|
| 65 |
+
"full_attention",
|
| 66 |
"full_attention"
|
| 67 |
],
|
| 68 |
"max_position_embeddings": 8192,
|
| 69 |
+
"max_window_layers": 48,
|
| 70 |
+
"mlp_type": "squared_relu",
|
| 71 |
"model_name": "test_checkpoint",
|
| 72 |
"model_type": "qwen3",
|
| 73 |
+
"n_layer": 48,
|
| 74 |
"num_attention_heads": 16,
|
| 75 |
+
"num_hidden_layers": 48,
|
| 76 |
"num_key_value_heads": 4,
|
| 77 |
"pad_token_id": 50034,
|
| 78 |
+
"resid_lambda_end": 1.05,
|
| 79 |
+
"resid_lambda_init": 1.15,
|
| 80 |
+
"resid_lambda_init_end": 1.05,
|
| 81 |
+
"resid_lambda_init_start": 1.15,
|
| 82 |
+
"resid_scalar_lr_mult": 0.01,
|
| 83 |
+
"resid_scalar_weight_decay": 0.05,
|
| 84 |
"rms_norm_eps": 1e-06,
|
| 85 |
"rope_parameters": {
|
| 86 |
"rope_theta": 50000,
|
| 87 |
"rope_type": "default"
|
| 88 |
},
|
| 89 |
+
"scalar_lr": 0.5,
|
| 90 |
"sliding_window": null,
|
| 91 |
+
"squared_relu_activation": "relu2",
|
| 92 |
+
"squared_relu_intermediate_size": 2880,
|
| 93 |
+
"tie_word_embeddings": false,
|
|
|
|
|
|
|
| 94 |
"transformers_version": "5.7.0",
|
| 95 |
"unsloth_version": "2026.4.8",
|
| 96 |
"use_cache": false,
|
|
|
|
| 97 |
"use_sliding_window": false,
|
| 98 |
"vocab_size": 50048,
|
| 99 |
+
"x0_lambda_end": 0.0,
|
| 100 |
+
"x0_lambda_init": 0.02,
|
| 101 |
+
"x0_lambda_init_end": 0.05,
|
| 102 |
+
"x0_lambda_init_start": 0.2,
|
| 103 |
+
"x0_scalar_weight_decay": 0.0
|
| 104 |
}
|