task-welfare-selection: m_models.tsv
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- 12b/V_bottom-s0/README.md +206 -0
- 12b/V_bottom-s0/adapter_config.json +186 -0
- 12b/V_bottom-s0/adapter_model.safetensors +3 -0
- 12b/V_bottom-s0/config.json +33 -0
- 12b/V_bottom-s0/log.jsonl +312 -0
- 12b/V_top-s0/README.md +206 -0
- 12b/V_top-s0/adapter_config.json +186 -0
- 12b/V_top-s0/adapter_model.safetensors +3 -0
- 12b/V_top-s0/config.json +33 -0
- 12b/V_top-s0/log.jsonl +312 -0
- 12b/u_bottom-s0/README.md +206 -0
- 12b/u_bottom-s0/adapter_config.json +186 -0
- 12b/u_bottom-s0/adapter_model.safetensors +3 -0
- 12b/u_bottom-s0/config.json +33 -0
- 12b/u_bottom-s0/log.jsonl +312 -0
- 12b/u_top-s0/README.md +206 -0
- 12b/u_top-s0/adapter_config.json +186 -0
- 12b/u_top-s0/adapter_model.safetensors +3 -0
- 12b/u_top-s0/config.json +33 -0
- 12b/u_top-s0/log.jsonl +312 -0
- README.md +123 -0
- gemma-3-27b-it/bottom-s0/README.md +206 -0
- gemma-3-27b-it/bottom-s0/adapter_config.json +228 -0
- gemma-3-27b-it/bottom-s0/adapter_model.safetensors +3 -0
- gemma-3-27b-it/bottom-s0/config.json +33 -0
- gemma-3-27b-it/bottom-s0/log.jsonl +143 -0
- gemma-3-27b-it/haiku_bottom-s0/README.md +206 -0
- gemma-3-27b-it/haiku_bottom-s0/adapter_config.json +228 -0
- gemma-3-27b-it/haiku_bottom-s0/adapter_model.safetensors +3 -0
- gemma-3-27b-it/haiku_bottom-s0/config.json +33 -0
- gemma-3-27b-it/haiku_bottom-s0/log.jsonl +312 -0
- gemma-3-27b-it/haiku_top-s0/README.md +206 -0
- gemma-3-27b-it/haiku_top-s0/adapter_config.json +228 -0
- gemma-3-27b-it/haiku_top-s0/adapter_model.safetensors +3 -0
- gemma-3-27b-it/haiku_top-s0/config.json +33 -0
- gemma-3-27b-it/haiku_top-s0/log.jsonl +312 -0
- gemma-3-27b-it/top-s0/README.md +206 -0
- gemma-3-27b-it/top-s0/adapter_config.json +228 -0
- gemma-3-27b-it/top-s0/adapter_model.safetensors +3 -0
- gemma-3-27b-it/top-s0/config.json +33 -0
- gemma-3-27b-it/top-s0/log.jsonl +143 -0
- qwen3-14b/bottom-s0/README.md +206 -0
- qwen3-14b/bottom-s0/adapter_config.json +45 -0
- qwen3-14b/bottom-s0/adapter_model.safetensors +3 -0
- qwen3-14b/bottom-s0/config.json +34 -0
- qwen3-14b/bottom-s0/log.jsonl +296 -0
- qwen3-14b/rand-s0/README.md +206 -0
- qwen3-14b/rand-s0/adapter_config.json +45 -0
- qwen3-14b/rand-s0/adapter_model.safetensors +3 -0
- qwen3-14b/rand-s0/config.json +34 -0
12b/V_bottom-s0/README.md
ADDED
|
@@ -0,0 +1,206 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: google/gemma-3-12b-it
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- base_model:adapter:google/gemma-3-12b-it
|
| 6 |
+
- lora
|
| 7 |
+
- transformers
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# Model Card for Model ID
|
| 11 |
+
|
| 12 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
## Model Details
|
| 17 |
+
|
| 18 |
+
### Model Description
|
| 19 |
+
|
| 20 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
- **Developed by:** [More Information Needed]
|
| 25 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 26 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 27 |
+
- **Model type:** [More Information Needed]
|
| 28 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 29 |
+
- **License:** [More Information Needed]
|
| 30 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 31 |
+
|
| 32 |
+
### Model Sources [optional]
|
| 33 |
+
|
| 34 |
+
<!-- Provide the basic links for the model. -->
|
| 35 |
+
|
| 36 |
+
- **Repository:** [More Information Needed]
|
| 37 |
+
- **Paper [optional]:** [More Information Needed]
|
| 38 |
+
- **Demo [optional]:** [More Information Needed]
|
| 39 |
+
|
| 40 |
+
## Uses
|
| 41 |
+
|
| 42 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 43 |
+
|
| 44 |
+
### Direct Use
|
| 45 |
+
|
| 46 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 47 |
+
|
| 48 |
+
[More Information Needed]
|
| 49 |
+
|
| 50 |
+
### Downstream Use [optional]
|
| 51 |
+
|
| 52 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 53 |
+
|
| 54 |
+
[More Information Needed]
|
| 55 |
+
|
| 56 |
+
### Out-of-Scope Use
|
| 57 |
+
|
| 58 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 59 |
+
|
| 60 |
+
[More Information Needed]
|
| 61 |
+
|
| 62 |
+
## Bias, Risks, and Limitations
|
| 63 |
+
|
| 64 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 65 |
+
|
| 66 |
+
[More Information Needed]
|
| 67 |
+
|
| 68 |
+
### Recommendations
|
| 69 |
+
|
| 70 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 71 |
+
|
| 72 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 73 |
+
|
| 74 |
+
## How to Get Started with the Model
|
| 75 |
+
|
| 76 |
+
Use the code below to get started with the model.
|
| 77 |
+
|
| 78 |
+
[More Information Needed]
|
| 79 |
+
|
| 80 |
+
## Training Details
|
| 81 |
+
|
| 82 |
+
### Training Data
|
| 83 |
+
|
| 84 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 85 |
+
|
| 86 |
+
[More Information Needed]
|
| 87 |
+
|
| 88 |
+
### Training Procedure
|
| 89 |
+
|
| 90 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 91 |
+
|
| 92 |
+
#### Preprocessing [optional]
|
| 93 |
+
|
| 94 |
+
[More Information Needed]
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
#### Training Hyperparameters
|
| 98 |
+
|
| 99 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 100 |
+
|
| 101 |
+
#### Speeds, Sizes, Times [optional]
|
| 102 |
+
|
| 103 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 104 |
+
|
| 105 |
+
[More Information Needed]
|
| 106 |
+
|
| 107 |
+
## Evaluation
|
| 108 |
+
|
| 109 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 110 |
+
|
| 111 |
+
### Testing Data, Factors & Metrics
|
| 112 |
+
|
| 113 |
+
#### Testing Data
|
| 114 |
+
|
| 115 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 116 |
+
|
| 117 |
+
[More Information Needed]
|
| 118 |
+
|
| 119 |
+
#### Factors
|
| 120 |
+
|
| 121 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 122 |
+
|
| 123 |
+
[More Information Needed]
|
| 124 |
+
|
| 125 |
+
#### Metrics
|
| 126 |
+
|
| 127 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 128 |
+
|
| 129 |
+
[More Information Needed]
|
| 130 |
+
|
| 131 |
+
### Results
|
| 132 |
+
|
| 133 |
+
[More Information Needed]
|
| 134 |
+
|
| 135 |
+
#### Summary
|
| 136 |
+
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
## Model Examination [optional]
|
| 140 |
+
|
| 141 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 142 |
+
|
| 143 |
+
[More Information Needed]
|
| 144 |
+
|
| 145 |
+
## Environmental Impact
|
| 146 |
+
|
| 147 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 148 |
+
|
| 149 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 150 |
+
|
| 151 |
+
- **Hardware Type:** [More Information Needed]
|
| 152 |
+
- **Hours used:** [More Information Needed]
|
| 153 |
+
- **Cloud Provider:** [More Information Needed]
|
| 154 |
+
- **Compute Region:** [More Information Needed]
|
| 155 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 156 |
+
|
| 157 |
+
## Technical Specifications [optional]
|
| 158 |
+
|
| 159 |
+
### Model Architecture and Objective
|
| 160 |
+
|
| 161 |
+
[More Information Needed]
|
| 162 |
+
|
| 163 |
+
### Compute Infrastructure
|
| 164 |
+
|
| 165 |
+
[More Information Needed]
|
| 166 |
+
|
| 167 |
+
#### Hardware
|
| 168 |
+
|
| 169 |
+
[More Information Needed]
|
| 170 |
+
|
| 171 |
+
#### Software
|
| 172 |
+
|
| 173 |
+
[More Information Needed]
|
| 174 |
+
|
| 175 |
+
## Citation [optional]
|
| 176 |
+
|
| 177 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 178 |
+
|
| 179 |
+
**BibTeX:**
|
| 180 |
+
|
| 181 |
+
[More Information Needed]
|
| 182 |
+
|
| 183 |
+
**APA:**
|
| 184 |
+
|
| 185 |
+
[More Information Needed]
|
| 186 |
+
|
| 187 |
+
## Glossary [optional]
|
| 188 |
+
|
| 189 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 190 |
+
|
| 191 |
+
[More Information Needed]
|
| 192 |
+
|
| 193 |
+
## More Information [optional]
|
| 194 |
+
|
| 195 |
+
[More Information Needed]
|
| 196 |
+
|
| 197 |
+
## Model Card Authors [optional]
|
| 198 |
+
|
| 199 |
+
[More Information Needed]
|
| 200 |
+
|
| 201 |
+
## Model Card Contact
|
| 202 |
+
|
| 203 |
+
[More Information Needed]
|
| 204 |
+
### Framework versions
|
| 205 |
+
|
| 206 |
+
- PEFT 0.17.1
|
12b/V_bottom-s0/adapter_config.json
ADDED
|
@@ -0,0 +1,186 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": {
|
| 4 |
+
"base_model_class": "Gemma3ForConditionalGeneration",
|
| 5 |
+
"parent_library": "transformers.models.gemma3.modeling_gemma3"
|
| 6 |
+
},
|
| 7 |
+
"base_model_name_or_path": "google/gemma-3-12b-it",
|
| 8 |
+
"bias": "none",
|
| 9 |
+
"corda_config": null,
|
| 10 |
+
"eva_config": null,
|
| 11 |
+
"exclude_modules": null,
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": true,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layer_replication": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"layers_to_transform": null,
|
| 18 |
+
"loftq_config": {},
|
| 19 |
+
"lora_alpha": 128,
|
| 20 |
+
"lora_bias": false,
|
| 21 |
+
"lora_dropout": 0.05,
|
| 22 |
+
"megatron_config": null,
|
| 23 |
+
"megatron_core": "megatron.core",
|
| 24 |
+
"modules_to_save": null,
|
| 25 |
+
"peft_type": "LORA",
|
| 26 |
+
"qalora_group_size": 16,
|
| 27 |
+
"r": 64,
|
| 28 |
+
"rank_pattern": {},
|
| 29 |
+
"revision": null,
|
| 30 |
+
"target_modules": [
|
| 31 |
+
"language_model.layers.6.self_attn.v_proj",
|
| 32 |
+
"language_model.layers.18.self_attn.v_proj",
|
| 33 |
+
"34.self_attn.q_proj",
|
| 34 |
+
"37.self_attn.v_proj",
|
| 35 |
+
"language_model.layers.25.self_attn.k_proj",
|
| 36 |
+
"language_model.layers.2.self_attn.k_proj",
|
| 37 |
+
"language_model.layers.21.self_attn.v_proj",
|
| 38 |
+
"45.self_attn.v_proj",
|
| 39 |
+
"language_model.layers.26.self_attn.k_proj",
|
| 40 |
+
"language_model.layers.24.self_attn.v_proj",
|
| 41 |
+
"language_model.layers.9.self_attn.k_proj",
|
| 42 |
+
"34.self_attn.v_proj",
|
| 43 |
+
"41.self_attn.k_proj",
|
| 44 |
+
"31.self_attn.q_proj",
|
| 45 |
+
"36.self_attn.v_proj",
|
| 46 |
+
"29.self_attn.k_proj",
|
| 47 |
+
"42.self_attn.v_proj",
|
| 48 |
+
"27.self_attn.q_proj",
|
| 49 |
+
"language_model.layers.26.self_attn.q_proj",
|
| 50 |
+
"language_model.layers.14.self_attn.k_proj",
|
| 51 |
+
"language_model.layers.25.self_attn.v_proj",
|
| 52 |
+
"language_model.layers.4.self_attn.q_proj",
|
| 53 |
+
"language_model.layers.22.self_attn.q_proj",
|
| 54 |
+
"language_model.layers.8.self_attn.v_proj",
|
| 55 |
+
"language_model.layers.11.self_attn.q_proj",
|
| 56 |
+
"28.self_attn.v_proj",
|
| 57 |
+
"40.self_attn.q_proj",
|
| 58 |
+
"up_proj",
|
| 59 |
+
"37.self_attn.k_proj",
|
| 60 |
+
"language_model.layers.2.self_attn.v_proj",
|
| 61 |
+
"language_model.layers.22.self_attn.k_proj",
|
| 62 |
+
"language_model.layers.7.self_attn.v_proj",
|
| 63 |
+
"27.self_attn.v_proj",
|
| 64 |
+
"language_model.layers.10.self_attn.v_proj",
|
| 65 |
+
"language_model.layers.13.self_attn.v_proj",
|
| 66 |
+
"language_model.layers.5.self_attn.v_proj",
|
| 67 |
+
"43.self_attn.v_proj",
|
| 68 |
+
"language_model.layers.3.self_attn.k_proj",
|
| 69 |
+
"language_model.layers.6.self_attn.q_proj",
|
| 70 |
+
"31.self_attn.v_proj",
|
| 71 |
+
"40.self_attn.k_proj",
|
| 72 |
+
"33.self_attn.k_proj",
|
| 73 |
+
"language_model.layers.19.self_attn.k_proj",
|
| 74 |
+
"language_model.layers.12.self_attn.k_proj",
|
| 75 |
+
"45.self_attn.k_proj",
|
| 76 |
+
"language_model.layers.18.self_attn.q_proj",
|
| 77 |
+
"language_model.layers.16.self_attn.k_proj",
|
| 78 |
+
"gate_proj",
|
| 79 |
+
"language_model.layers.3.self_attn.q_proj",
|
| 80 |
+
"43.self_attn.k_proj",
|
| 81 |
+
"language_model.layers.15.self_attn.k_proj",
|
| 82 |
+
"language_model.layers.20.self_attn.v_proj",
|
| 83 |
+
"language_model.layers.23.self_attn.q_proj",
|
| 84 |
+
"44.self_attn.v_proj",
|
| 85 |
+
"31.self_attn.k_proj",
|
| 86 |
+
"language_model.layers.2.self_attn.q_proj",
|
| 87 |
+
"language_model.layers.3.self_attn.v_proj",
|
| 88 |
+
"44.self_attn.k_proj",
|
| 89 |
+
"language_model.layers.11.self_attn.k_proj",
|
| 90 |
+
"32.self_attn.v_proj",
|
| 91 |
+
"41.self_attn.v_proj",
|
| 92 |
+
"down_proj",
|
| 93 |
+
"language_model.layers.1.self_attn.k_proj",
|
| 94 |
+
"language_model.layers.25.self_attn.q_proj",
|
| 95 |
+
"29.self_attn.q_proj",
|
| 96 |
+
"42.self_attn.q_proj",
|
| 97 |
+
"o_proj",
|
| 98 |
+
"35.self_attn.q_proj",
|
| 99 |
+
"language_model.layers.26.self_attn.v_proj",
|
| 100 |
+
"language_model.layers.24.self_attn.k_proj",
|
| 101 |
+
"language_model.layers.24.self_attn.q_proj",
|
| 102 |
+
"47.self_attn.q_proj",
|
| 103 |
+
"27.self_attn.k_proj",
|
| 104 |
+
"language_model.layers.17.self_attn.q_proj",
|
| 105 |
+
"language_model.layers.20.self_attn.q_proj",
|
| 106 |
+
"33.self_attn.q_proj",
|
| 107 |
+
"39.self_attn.q_proj",
|
| 108 |
+
"language_model.layers.14.self_attn.q_proj",
|
| 109 |
+
"language_model.layers.16.self_attn.q_proj",
|
| 110 |
+
"47.self_attn.v_proj",
|
| 111 |
+
"language_model.layers.1.self_attn.v_proj",
|
| 112 |
+
"36.self_attn.q_proj",
|
| 113 |
+
"42.self_attn.k_proj",
|
| 114 |
+
"language_model.layers.15.self_attn.q_proj",
|
| 115 |
+
"45.self_attn.q_proj",
|
| 116 |
+
"29.self_attn.v_proj",
|
| 117 |
+
"language_model.layers.10.self_attn.q_proj",
|
| 118 |
+
"language_model.layers.12.self_attn.v_proj",
|
| 119 |
+
"35.self_attn.v_proj",
|
| 120 |
+
"30.self_attn.v_proj",
|
| 121 |
+
"44.self_attn.q_proj",
|
| 122 |
+
"41.self_attn.q_proj",
|
| 123 |
+
"39.self_attn.v_proj",
|
| 124 |
+
"38.self_attn.q_proj",
|
| 125 |
+
"language_model.layers.15.self_attn.v_proj",
|
| 126 |
+
"language_model.layers.9.self_attn.q_proj",
|
| 127 |
+
"language_model.layers.18.self_attn.k_proj",
|
| 128 |
+
"language_model.layers.6.self_attn.k_proj",
|
| 129 |
+
"32.self_attn.k_proj",
|
| 130 |
+
"35.self_attn.k_proj",
|
| 131 |
+
"40.self_attn.v_proj",
|
| 132 |
+
"47.self_attn.k_proj",
|
| 133 |
+
"32.self_attn.q_proj",
|
| 134 |
+
"43.self_attn.q_proj",
|
| 135 |
+
"language_model.layers.4.self_attn.k_proj",
|
| 136 |
+
"language_model.layers.7.self_attn.q_proj",
|
| 137 |
+
"30.self_attn.k_proj",
|
| 138 |
+
"38.self_attn.k_proj",
|
| 139 |
+
"language_model.layers.11.self_attn.v_proj",
|
| 140 |
+
"28.self_attn.k_proj",
|
| 141 |
+
"language_model.layers.1.self_attn.q_proj",
|
| 142 |
+
"46.self_attn.k_proj",
|
| 143 |
+
"language_model.layers.19.self_attn.q_proj",
|
| 144 |
+
"39.self_attn.k_proj",
|
| 145 |
+
"language_model.layers.12.self_attn.q_proj",
|
| 146 |
+
"34.self_attn.k_proj",
|
| 147 |
+
"language_model.layers.22.self_attn.v_proj",
|
| 148 |
+
"language_model.layers.21.self_attn.k_proj",
|
| 149 |
+
"language_model.layers.21.self_attn.q_proj",
|
| 150 |
+
"language_model.layers.14.self_attn.v_proj",
|
| 151 |
+
"language_model.layers.16.self_attn.v_proj",
|
| 152 |
+
"language_model.layers.7.self_attn.k_proj",
|
| 153 |
+
"language_model.layers.0.self_attn.v_proj",
|
| 154 |
+
"language_model.layers.5.self_attn.k_proj",
|
| 155 |
+
"language_model.layers.23.self_attn.v_proj",
|
| 156 |
+
"language_model.layers.20.self_attn.k_proj",
|
| 157 |
+
"language_model.layers.19.self_attn.v_proj",
|
| 158 |
+
"36.self_attn.k_proj",
|
| 159 |
+
"language_model.layers.17.self_attn.v_proj",
|
| 160 |
+
"language_model.layers.13.self_attn.k_proj",
|
| 161 |
+
"language_model.layers.23.self_attn.k_proj",
|
| 162 |
+
"37.self_attn.q_proj",
|
| 163 |
+
"language_model.layers.4.self_attn.v_proj",
|
| 164 |
+
"language_model.layers.0.self_attn.q_proj",
|
| 165 |
+
"language_model.layers.17.self_attn.k_proj",
|
| 166 |
+
"language_model.layers.0.self_attn.k_proj",
|
| 167 |
+
"30.self_attn.q_proj",
|
| 168 |
+
"language_model.layers.5.self_attn.q_proj",
|
| 169 |
+
"language_model.layers.8.self_attn.q_proj",
|
| 170 |
+
"language_model.layers.13.self_attn.q_proj",
|
| 171 |
+
"46.self_attn.q_proj",
|
| 172 |
+
"language_model.layers.8.self_attn.k_proj",
|
| 173 |
+
"language_model.layers.9.self_attn.v_proj",
|
| 174 |
+
"language_model.layers.10.self_attn.k_proj",
|
| 175 |
+
"33.self_attn.v_proj",
|
| 176 |
+
"38.self_attn.v_proj",
|
| 177 |
+
"28.self_attn.q_proj",
|
| 178 |
+
"46.self_attn.v_proj"
|
| 179 |
+
],
|
| 180 |
+
"target_parameters": null,
|
| 181 |
+
"task_type": null,
|
| 182 |
+
"trainable_token_indices": null,
|
| 183 |
+
"use_dora": false,
|
| 184 |
+
"use_qalora": false,
|
| 185 |
+
"use_rslora": false
|
| 186 |
+
}
|
12b/V_bottom-s0/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c005ba534be480cf68ed8c2dfa9c9d148f539a72ff958c308fcc7c9274771893
|
| 3 |
+
size 1047628488
|
12b/V_bottom-s0/config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "google/gemma-3-12b-it",
|
| 3 |
+
"data": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma12/arms_v/TASKW-V_bottom-gemma12.jsonl",
|
| 4 |
+
"run": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma12/runs/V_bottom-s0",
|
| 5 |
+
"steps": 0,
|
| 6 |
+
"batch": 16,
|
| 7 |
+
"micro": 1,
|
| 8 |
+
"seq_len": 4096,
|
| 9 |
+
"lr": 0.0001,
|
| 10 |
+
"warmup": 10,
|
| 11 |
+
"lora_rank": 64,
|
| 12 |
+
"lora_targets": "all-linear",
|
| 13 |
+
"lora_targets_file": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/code/gemma3_12b_lora_targets.json",
|
| 14 |
+
"seed": 0,
|
| 15 |
+
"bos": true,
|
| 16 |
+
"no_merge": false,
|
| 17 |
+
"save_at": "",
|
| 18 |
+
"replay": "",
|
| 19 |
+
"replay_frac": 0.0,
|
| 20 |
+
"constant_lr": false,
|
| 21 |
+
"schedule": "hf-cosine",
|
| 22 |
+
"min_lr_frac": 0.0,
|
| 23 |
+
"warmup_ratio": 0.05,
|
| 24 |
+
"lora_dropout": 0.05,
|
| 25 |
+
"no_pack": true,
|
| 26 |
+
"chat_sft": true,
|
| 27 |
+
"data_frac": 1.0,
|
| 28 |
+
"base_adapter": "",
|
| 29 |
+
"ckpt_every": 0,
|
| 30 |
+
"kill_after": 0,
|
| 31 |
+
"method": "doc_lm_lora",
|
| 32 |
+
"world": 2
|
| 33 |
+
}
|
12b/V_bottom-s0/log.jsonl
ADDED
|
@@ -0,0 +1,312 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 1, "loss": 2.262601375579834, "grad_norm": 4.321106910705566, "lr": 1.25e-05, "epoch": 1, "t": 4.0903847217559814}
|
| 2 |
+
{"step": 2, "loss": 2.1671199798583984, "grad_norm": 3.742156982421875, "lr": 1.8750000000000002e-05, "epoch": 1, "t": 4.178858280181885}
|
| 3 |
+
{"step": 3, "loss": 2.0548484325408936, "grad_norm": 3.4029154777526855, "lr": 2.5e-05, "epoch": 1, "t": 3.7915890216827393}
|
| 4 |
+
{"step": 4, "loss": 1.9494054317474365, "grad_norm": 2.825042247772217, "lr": 3.125e-05, "epoch": 1, "t": 3.1633121967315674}
|
| 5 |
+
{"step": 5, "loss": 1.820056438446045, "grad_norm": 2.2459218502044678, "lr": 3.7500000000000003e-05, "epoch": 1, "t": 3.651564598083496}
|
| 6 |
+
{"step": 6, "loss": 1.7989606857299805, "grad_norm": 1.4922831058502197, "lr": 4.375e-05, "epoch": 1, "t": 3.106041193008423}
|
| 7 |
+
{"step": 7, "loss": 1.7327988147735596, "grad_norm": 1.0837712287902832, "lr": 5e-05, "epoch": 1, "t": 2.8009917736053467}
|
| 8 |
+
{"step": 8, "loss": 1.5018919706344604, "grad_norm": 0.8392522931098938, "lr": 5.6250000000000005e-05, "epoch": 1, "t": 3.020111322402954}
|
| 9 |
+
{"step": 9, "loss": 1.6870667934417725, "grad_norm": 0.9892560839653015, "lr": 6.25e-05, "epoch": 1, "t": 3.2018043994903564}
|
| 10 |
+
{"step": 10, "loss": 1.585772156715393, "grad_norm": 1.000664472579956, "lr": 6.875e-05, "epoch": 1, "t": 2.913628101348877}
|
| 11 |
+
{"step": 11, "loss": 1.6362735033035278, "grad_norm": 0.9887240529060364, "lr": 7.500000000000001e-05, "epoch": 1, "t": 3.0572726726531982}
|
| 12 |
+
{"step": 12, "loss": 1.41947340965271, "grad_norm": 0.8526573777198792, "lr": 8.125000000000001e-05, "epoch": 1, "t": 3.5328314304351807}
|
| 13 |
+
{"step": 13, "loss": 1.2581584453582764, "grad_norm": 0.7512729167938232, "lr": 8.75e-05, "epoch": 1, "t": 3.1377861499786377}
|
| 14 |
+
{"step": 14, "loss": 1.4302926063537598, "grad_norm": 0.779813826084137, "lr": 9.375e-05, "epoch": 1, "t": 2.6500909328460693}
|
| 15 |
+
{"step": 15, "loss": 1.5263147354125977, "grad_norm": 0.7511342763900757, "lr": 0.0001, "epoch": 1, "t": 2.988670825958252}
|
| 16 |
+
{"step": 16, "loss": 1.3866556882858276, "grad_norm": 0.7250912189483643, "lr": 0.0001, "epoch": 1, "t": 3.1635689735412598}
|
| 17 |
+
{"step": 17, "loss": 1.3464088439941406, "grad_norm": 0.7572988867759705, "lr": 9.99971838728789e-05, "epoch": 1, "t": 3.2264647483825684}
|
| 18 |
+
{"step": 18, "loss": 1.369933843612671, "grad_norm": 0.7048869132995605, "lr": 9.998873580873848e-05, "epoch": 1, "t": 2.8280229568481445}
|
| 19 |
+
{"step": 19, "loss": 1.3470475673675537, "grad_norm": 0.7550159096717834, "lr": 9.997465675921163e-05, "epoch": 1, "t": 2.953085422515869}
|
| 20 |
+
{"step": 20, "loss": 1.3588720560073853, "grad_norm": 0.6520718932151794, "lr": 9.995494831023409e-05, "epoch": 1, "t": 2.9637160301208496}
|
| 21 |
+
{"step": 21, "loss": 1.2363781929016113, "grad_norm": 0.608392596244812, "lr": 9.992961268186573e-05, "epoch": 1, "t": 3.0464730262756348}
|
| 22 |
+
{"step": 22, "loss": 1.3877007961273193, "grad_norm": 0.6800710558891296, "lr": 9.989865272804063e-05, "epoch": 1, "t": 2.89674973487854}
|
| 23 |
+
{"step": 23, "loss": 1.2164907455444336, "grad_norm": 0.592847466468811, "lr": 9.986207193624536e-05, "epoch": 1, "t": 3.050445318222046}
|
| 24 |
+
{"step": 24, "loss": 1.3283023834228516, "grad_norm": 0.5463132858276367, "lr": 9.981987442712633e-05, "epoch": 1, "t": 2.826953411102295}
|
| 25 |
+
{"step": 25, "loss": 1.1635342836380005, "grad_norm": 0.5890938639640808, "lr": 9.977206495402554e-05, "epoch": 1, "t": 2.8069756031036377}
|
| 26 |
+
{"step": 26, "loss": 1.2780532836914062, "grad_norm": 0.6457180380821228, "lr": 9.971864890244513e-05, "epoch": 1, "t": 3.0348150730133057}
|
| 27 |
+
{"step": 27, "loss": 1.1895959377288818, "grad_norm": 0.678081750869751, "lr": 9.965963228944078e-05, "epoch": 1, "t": 3.493077278137207}
|
| 28 |
+
{"step": 28, "loss": 1.2893505096435547, "grad_norm": 0.5789566040039062, "lr": 9.959502176294383e-05, "epoch": 1, "t": 3.206479072570801}
|
| 29 |
+
{"step": 29, "loss": 1.175737738609314, "grad_norm": 0.7599726915359497, "lr": 9.95248246010126e-05, "epoch": 1, "t": 2.775560140609741}
|
| 30 |
+
{"step": 30, "loss": 1.3553040027618408, "grad_norm": 0.5916820168495178, "lr": 9.944904871101228e-05, "epoch": 1, "t": 3.330934762954712}
|
| 31 |
+
{"step": 31, "loss": 1.339888334274292, "grad_norm": 0.6464534401893616, "lr": 9.936770262872443e-05, "epoch": 1, "t": 2.7135303020477295}
|
| 32 |
+
{"step": 32, "loss": 1.3124827146530151, "grad_norm": 0.6097561717033386, "lr": 9.928079551738543e-05, "epoch": 1, "t": 3.222599744796753}
|
| 33 |
+
{"step": 33, "loss": 1.298492670059204, "grad_norm": 0.627341091632843, "lr": 9.918833716665419e-05, "epoch": 1, "t": 2.9775819778442383}
|
| 34 |
+
{"step": 34, "loss": 1.2930155992507935, "grad_norm": 0.6147288680076599, "lr": 9.909033799150946e-05, "epoch": 1, "t": 3.3722410202026367}
|
| 35 |
+
{"step": 35, "loss": 1.3372199535369873, "grad_norm": 0.6341346502304077, "lr": 9.898680903107666e-05, "epoch": 1, "t": 2.9427356719970703}
|
| 36 |
+
{"step": 36, "loss": 1.2179431915283203, "grad_norm": 0.5810100436210632, "lr": 9.887776194738432e-05, "epoch": 1, "t": 3.1459813117980957}
|
| 37 |
+
{"step": 37, "loss": 1.2858481407165527, "grad_norm": 0.5698282122612, "lr": 9.876320902405042e-05, "epoch": 1, "t": 3.182222843170166}
|
| 38 |
+
{"step": 38, "loss": 1.1030207872390747, "grad_norm": 0.6010777354240417, "lr": 9.864316316489873e-05, "epoch": 1, "t": 2.9186089038848877}
|
| 39 |
+
{"step": 39, "loss": 1.1773898601531982, "grad_norm": 0.6125306487083435, "lr": 9.851763789250525e-05, "epoch": 1, "t": 3.0962703227996826}
|
| 40 |
+
{"step": 40, "loss": 1.2243480682373047, "grad_norm": 0.6106325387954712, "lr": 9.838664734667495e-05, "epoch": 1, "t": 2.9965648651123047}
|
| 41 |
+
{"step": 41, "loss": 1.241113543510437, "grad_norm": 0.556209921836853, "lr": 9.825020628284896e-05, "epoch": 1, "t": 2.77687931060791}
|
| 42 |
+
{"step": 42, "loss": 1.3011119365692139, "grad_norm": 0.6555788516998291, "lr": 9.810833007044247e-05, "epoch": 1, "t": 3.0246005058288574}
|
| 43 |
+
{"step": 43, "loss": 1.316701889038086, "grad_norm": 0.5527663230895996, "lr": 9.796103469111351e-05, "epoch": 1, "t": 3.6717283725738525}
|
| 44 |
+
{"step": 44, "loss": 1.2274091243743896, "grad_norm": 0.590487539768219, "lr": 9.780833673696254e-05, "epoch": 1, "t": 3.0330426692962646}
|
| 45 |
+
{"step": 45, "loss": 1.1903352737426758, "grad_norm": 0.5301953554153442, "lr": 9.76502534086636e-05, "epoch": 1, "t": 3.1038732528686523}
|
| 46 |
+
{"step": 46, "loss": 1.2604535818099976, "grad_norm": 0.6428079605102539, "lr": 9.74868025135266e-05, "epoch": 1, "t": 2.9908602237701416}
|
| 47 |
+
{"step": 47, "loss": 1.233270287513733, "grad_norm": 0.6347683668136597, "lr": 9.731800246349148e-05, "epoch": 1, "t": 3.0215275287628174}
|
| 48 |
+
{"step": 48, "loss": 1.2261725664138794, "grad_norm": 0.6844096779823303, "lr": 9.714387227305422e-05, "epoch": 1, "t": 3.475930690765381}
|
| 49 |
+
{"step": 49, "loss": 1.2167680263519287, "grad_norm": 0.5968977212905884, "lr": 9.696443155712486e-05, "epoch": 1, "t": 2.6926209926605225}
|
| 50 |
+
{"step": 50, "loss": 1.2470042705535889, "grad_norm": 0.6235132813453674, "lr": 9.67797005288181e-05, "epoch": 1, "t": 2.796698808670044}
|
| 51 |
+
{"step": 51, "loss": 1.270082950592041, "grad_norm": 0.5369529128074646, "lr": 9.65896999971763e-05, "epoch": 1, "t": 3.037039041519165}
|
| 52 |
+
{"step": 52, "loss": 1.176042079925537, "grad_norm": 0.5515763163566589, "lr": 9.639445136482548e-05, "epoch": 1, "t": 3.1277577877044678}
|
| 53 |
+
{"step": 53, "loss": 1.2403608560562134, "grad_norm": 0.6723449230194092, "lr": 9.619397662556435e-05, "epoch": 1, "t": 2.798184633255005}
|
| 54 |
+
{"step": 54, "loss": 1.3449089527130127, "grad_norm": 0.5897758603096008, "lr": 9.598829836188694e-05, "epoch": 1, "t": 3.058833122253418}
|
| 55 |
+
{"step": 55, "loss": 1.3472912311553955, "grad_norm": 0.6036477088928223, "lr": 9.577743974243874e-05, "epoch": 1, "t": 3.059793472290039}
|
| 56 |
+
{"step": 56, "loss": 1.205803394317627, "grad_norm": 0.5999453067779541, "lr": 9.55614245194068e-05, "epoch": 1, "t": 3.341444253921509}
|
| 57 |
+
{"step": 57, "loss": 1.2803912162780762, "grad_norm": 0.6849067211151123, "lr": 9.534027702584425e-05, "epoch": 1, "t": 3.402968406677246}
|
| 58 |
+
{"step": 58, "loss": 1.1960639953613281, "grad_norm": 0.5894226431846619, "lr": 9.511402217292926e-05, "epoch": 1, "t": 3.6668052673339844}
|
| 59 |
+
{"step": 59, "loss": 1.2295646667480469, "grad_norm": 0.61738520860672, "lr": 9.488268544715896e-05, "epoch": 1, "t": 3.121495008468628}
|
| 60 |
+
{"step": 60, "loss": 1.18776535987854, "grad_norm": 0.5862890481948853, "lr": 9.464629290747842e-05, "epoch": 1, "t": 3.8103444576263428}
|
| 61 |
+
{"step": 61, "loss": 1.34743332862854, "grad_norm": 0.548228919506073, "lr": 9.440487118234535e-05, "epoch": 1, "t": 3.465630054473877}
|
| 62 |
+
{"step": 62, "loss": 1.3217475414276123, "grad_norm": 0.5924105644226074, "lr": 9.415844746673047e-05, "epoch": 1, "t": 2.7527828216552734}
|
| 63 |
+
{"step": 63, "loss": 1.196764349937439, "grad_norm": 0.5375847220420837, "lr": 9.390704951905411e-05, "epoch": 1, "t": 3.2803151607513428}
|
| 64 |
+
{"step": 64, "loss": 1.3148972988128662, "grad_norm": 0.590986430644989, "lr": 9.365070565805941e-05, "epoch": 1, "t": 3.2135140895843506}
|
| 65 |
+
{"step": 65, "loss": 1.184212565422058, "grad_norm": 0.5995739698410034, "lr": 9.338944475962237e-05, "epoch": 1, "t": 2.9850845336914062}
|
| 66 |
+
{"step": 66, "loss": 1.1817116737365723, "grad_norm": 0.6395159363746643, "lr": 9.312329625349902e-05, "epoch": 1, "t": 2.839395046234131}
|
| 67 |
+
{"step": 67, "loss": 1.275105595588684, "grad_norm": 0.5798230171203613, "lr": 9.285229012001047e-05, "epoch": 1, "t": 3.1961710453033447}
|
| 68 |
+
{"step": 68, "loss": 1.2389097213745117, "grad_norm": 0.5572142601013184, "lr": 9.257645688666556e-05, "epoch": 1, "t": 3.0107476711273193}
|
| 69 |
+
{"step": 69, "loss": 1.1718604564666748, "grad_norm": 0.5847182869911194, "lr": 9.22958276247223e-05, "epoch": 1, "t": 2.824526309967041}
|
| 70 |
+
{"step": 70, "loss": 1.2704832553863525, "grad_norm": 0.6305415630340576, "lr": 9.201043394568773e-05, "epoch": 1, "t": 2.7763094902038574}
|
| 71 |
+
{"step": 71, "loss": 1.0497395992279053, "grad_norm": 0.5686884522438049, "lr": 9.172030799775699e-05, "epoch": 1, "t": 3.05053448677063}
|
| 72 |
+
{"step": 72, "loss": 1.2370097637176514, "grad_norm": 0.5335482954978943, "lr": 9.142548246219212e-05, "epoch": 1, "t": 2.962338924407959}
|
| 73 |
+
{"step": 73, "loss": 1.117645025253296, "grad_norm": 0.6910924911499023, "lr": 9.112599054964057e-05, "epoch": 1, "t": 2.7931530475616455}
|
| 74 |
+
{"step": 74, "loss": 1.1665549278259277, "grad_norm": 0.6299133896827698, "lr": 9.082186599639428e-05, "epoch": 1, "t": 2.966653347015381}
|
| 75 |
+
{"step": 75, "loss": 1.1305136680603027, "grad_norm": 0.5654453635215759, "lr": 9.051314306058933e-05, "epoch": 1, "t": 3.197213888168335}
|
| 76 |
+
{"step": 76, "loss": 1.1932015419006348, "grad_norm": 0.6124429702758789, "lr": 9.019985651834703e-05, "epoch": 1, "t": 3.2439706325531006}
|
| 77 |
+
{"step": 77, "loss": 1.2272367477416992, "grad_norm": 0.5745668411254883, "lr": 8.988204165985649e-05, "epoch": 1, "t": 3.235095977783203}
|
| 78 |
+
{"step": 78, "loss": 1.1715152263641357, "grad_norm": 0.5699538588523865, "lr": 8.955973428539944e-05, "epoch": 1, "t": 3.0144760608673096}
|
| 79 |
+
{"step": 79, "loss": 1.2684156894683838, "grad_norm": 0.5542939305305481, "lr": 8.923297070131737e-05, "epoch": 1, "t": 3.4508779048919678}
|
| 80 |
+
{"step": 80, "loss": 1.2495590448379517, "grad_norm": 0.6281068325042725, "lr": 8.890178771592199e-05, "epoch": 1, "t": 2.865173578262329}
|
| 81 |
+
{"step": 81, "loss": 1.2565178871154785, "grad_norm": 0.6397746205329895, "lr": 8.856622263534875e-05, "epoch": 1, "t": 2.889617681503296}
|
| 82 |
+
{"step": 82, "loss": 1.3881675004959106, "grad_norm": 0.6775103211402893, "lr": 8.822631325935463e-05, "epoch": 1, "t": 2.7949724197387695}
|
| 83 |
+
{"step": 83, "loss": 1.288458228111267, "grad_norm": 0.6452056169509888, "lr": 8.788209787706015e-05, "epoch": 1, "t": 2.9084620475769043}
|
| 84 |
+
{"step": 84, "loss": 1.2155323028564453, "grad_norm": 0.68458092212677, "lr": 8.753361526263621e-05, "epoch": 1, "t": 2.9407711029052734}
|
| 85 |
+
{"step": 85, "loss": 1.0869160890579224, "grad_norm": 0.7476386427879333, "lr": 8.718090467093654e-05, "epoch": 1, "t": 2.676293134689331}
|
| 86 |
+
{"step": 86, "loss": 1.143362283706665, "grad_norm": 0.5448579788208008, "lr": 8.682400583307562e-05, "epoch": 1, "t": 3.0713372230529785}
|
| 87 |
+
{"step": 87, "loss": 1.3163834810256958, "grad_norm": 0.5694847106933594, "lr": 8.646295895195333e-05, "epoch": 1, "t": 3.127225160598755}
|
| 88 |
+
{"step": 88, "loss": 1.0603773593902588, "grad_norm": 0.6071080565452576, "lr": 8.609780469772623e-05, "epoch": 1, "t": 2.7289798259735107}
|
| 89 |
+
{"step": 89, "loss": 1.2000758647918701, "grad_norm": 0.5982691645622253, "lr": 8.572858420322627e-05, "epoch": 1, "t": 2.9588217735290527}
|
| 90 |
+
{"step": 90, "loss": 1.2668378353118896, "grad_norm": 0.545132577419281, "lr": 8.535533905932738e-05, "epoch": 1, "t": 3.0684101581573486}
|
| 91 |
+
{"step": 91, "loss": 1.2654061317443848, "grad_norm": 0.6144419312477112, "lr": 8.497811131026046e-05, "epoch": 1, "t": 3.349602222442627}
|
| 92 |
+
{"step": 92, "loss": 1.1988807916641235, "grad_norm": 0.547225832939148, "lr": 8.459694344887732e-05, "epoch": 1, "t": 3.3556289672851562}
|
| 93 |
+
{"step": 93, "loss": 1.1566147804260254, "grad_norm": 0.6398140788078308, "lr": 8.421187841186402e-05, "epoch": 1, "t": 2.721677780151367}
|
| 94 |
+
{"step": 94, "loss": 1.293168306350708, "grad_norm": 0.6214326024055481, "lr": 8.382295957490436e-05, "epoch": 1, "t": 3.053834915161133}
|
| 95 |
+
{"step": 95, "loss": 1.2736681699752808, "grad_norm": 1.1022288799285889, "lr": 8.343023074779368e-05, "epoch": 1, "t": 3.348402261734009}
|
| 96 |
+
{"step": 96, "loss": 1.213397741317749, "grad_norm": 0.6194725036621094, "lr": 8.303373616950408e-05, "epoch": 1, "t": 2.614722490310669}
|
| 97 |
+
{"step": 97, "loss": 1.156456708908081, "grad_norm": 0.619104266166687, "lr": 8.263352050320094e-05, "epoch": 1, "t": 3.1172099113464355}
|
| 98 |
+
{"step": 98, "loss": 1.16475510597229, "grad_norm": 0.6148302555084229, "lr": 8.222962883121196e-05, "epoch": 1, "t": 2.827390670776367}
|
| 99 |
+
{"step": 99, "loss": 1.2601840496063232, "grad_norm": 0.6317527294158936, "lr": 8.182210664994878e-05, "epoch": 1, "t": 2.9053473472595215}
|
| 100 |
+
{"step": 100, "loss": 1.2153723239898682, "grad_norm": 0.6083713173866272, "lr": 8.141099986478212e-05, "epoch": 1, "t": 2.8607513904571533}
|
| 101 |
+
{"step": 101, "loss": 1.1575634479522705, "grad_norm": 0.880901575088501, "lr": 8.099635478487064e-05, "epoch": 1, "t": 3.005492687225342}
|
| 102 |
+
{"step": 102, "loss": 1.15977942943573, "grad_norm": 0.5449019074440002, "lr": 8.057821811794458e-05, "epoch": 1, "t": 3.3434953689575195}
|
| 103 |
+
{"step": 103, "loss": 1.3317471742630005, "grad_norm": 0.6378929018974304, "lr": 8.015663696504422e-05, "epoch": 1, "t": 2.9116969108581543}
|
| 104 |
+
{"step": 104, "loss": 1.2235063314437866, "grad_norm": 0.555363118648529, "lr": 7.973165881521434e-05, "epoch": 1, "t": 3.491269111633301}
|
| 105 |
+
{"step": 105, "loss": 1.1460270881652832, "grad_norm": 0.6242833137512207, "lr": 7.930333154015466e-05, "epoch": 1, "t": 3.0875916481018066}
|
| 106 |
+
{"step": 106, "loss": 1.2006102800369263, "grad_norm": 0.5549394488334656, "lr": 7.88717033888274e-05, "epoch": 1, "t": 3.341688394546509}
|
| 107 |
+
{"step": 107, "loss": 1.1297674179077148, "grad_norm": 0.6531938910484314, "lr": 7.843682298202235e-05, "epoch": 1, "t": 2.831692934036255}
|
| 108 |
+
{"step": 108, "loss": 1.2178858518600464, "grad_norm": 0.5434705018997192, "lr": 7.799873930687978e-05, "epoch": 1, "t": 3.7227730751037598}
|
| 109 |
+
{"step": 109, "loss": 1.288986086845398, "grad_norm": 0.5934914946556091, "lr": 7.755750171137246e-05, "epoch": 1, "t": 3.7522521018981934}
|
| 110 |
+
{"step": 110, "loss": 1.2648894786834717, "grad_norm": 0.6413567662239075, "lr": 7.711315989874677e-05, "epoch": 1, "t": 3.1471829414367676}
|
| 111 |
+
{"step": 111, "loss": 1.2437365055084229, "grad_norm": 0.6522145867347717, "lr": 7.666576392192389e-05, "epoch": 1, "t": 2.8307108879089355}
|
| 112 |
+
{"step": 112, "loss": 1.1486939191818237, "grad_norm": 0.6480277180671692, "lr": 7.621536417786159e-05, "epoch": 1, "t": 2.9968698024749756}
|
| 113 |
+
{"step": 113, "loss": 1.2442046403884888, "grad_norm": 0.6971429586410522, "lr": 7.576201140187727e-05, "epoch": 1, "t": 3.0079429149627686}
|
| 114 |
+
{"step": 114, "loss": 1.1534336805343628, "grad_norm": 0.5909664034843445, "lr": 7.530575666193283e-05, "epoch": 1, "t": 3.0501341819763184}
|
| 115 |
+
{"step": 115, "loss": 1.1915488243103027, "grad_norm": 0.633232593536377, "lr": 7.484665135288213e-05, "epoch": 1, "t": 2.769033670425415}
|
| 116 |
+
{"step": 116, "loss": 1.0821030139923096, "grad_norm": 0.7203329205513, "lr": 7.438474719068173e-05, "epoch": 1, "t": 3.0029449462890625}
|
| 117 |
+
{"step": 117, "loss": 1.2057216167449951, "grad_norm": 0.6815724968910217, "lr": 7.392009620656513e-05, "epoch": 1, "t": 2.8152177333831787}
|
| 118 |
+
{"step": 118, "loss": 1.1797432899475098, "grad_norm": 0.6875385046005249, "lr": 7.345275074118185e-05, "epoch": 1, "t": 2.7755990028381348}
|
| 119 |
+
{"step": 119, "loss": 1.1533446311950684, "grad_norm": 0.5891608595848083, "lr": 7.298276343870151e-05, "epoch": 1, "t": 3.287069320678711}
|
| 120 |
+
{"step": 120, "loss": 1.071908712387085, "grad_norm": 0.5524219274520874, "lr": 7.251018724088367e-05, "epoch": 1, "t": 3.0004425048828125}
|
| 121 |
+
{"step": 121, "loss": 1.1303341388702393, "grad_norm": 0.6244798302650452, "lr": 7.203507538111423e-05, "epoch": 1, "t": 2.8286290168762207}
|
| 122 |
+
{"step": 122, "loss": 1.1586410999298096, "grad_norm": 0.6407503485679626, "lr": 7.155748137840892e-05, "epoch": 1, "t": 2.9526290893554688}
|
| 123 |
+
{"step": 123, "loss": 1.3539068698883057, "grad_norm": 0.5737888813018799, "lr": 7.107745903138472e-05, "epoch": 1, "t": 3.1938977241516113}
|
| 124 |
+
{"step": 124, "loss": 1.3081499338150024, "grad_norm": 0.5784101486206055, "lr": 7.059506241219965e-05, "epoch": 1, "t": 3.176943063735962}
|
| 125 |
+
{"step": 125, "loss": 1.1669530868530273, "grad_norm": 0.5705282092094421, "lr": 7.011034586046176e-05, "epoch": 1, "t": 3.1705708503723145}
|
| 126 |
+
{"step": 126, "loss": 1.1863765716552734, "grad_norm": 0.5676631331443787, "lr": 6.962336397710819e-05, "epoch": 1, "t": 3.0866763591766357}
|
| 127 |
+
{"step": 127, "loss": 1.1052277088165283, "grad_norm": 0.6427485942840576, "lr": 6.91341716182545e-05, "epoch": 1, "t": 3.5048844814300537}
|
| 128 |
+
{"step": 128, "loss": 1.1741467714309692, "grad_norm": 0.6330947279930115, "lr": 6.864282388901544e-05, "epoch": 1, "t": 3.076354742050171}
|
| 129 |
+
{"step": 129, "loss": 1.0960209369659424, "grad_norm": 0.6368241906166077, "lr": 6.814937613729766e-05, "epoch": 1, "t": 2.942425012588501}
|
| 130 |
+
{"step": 130, "loss": 1.0821714401245117, "grad_norm": 0.6528913974761963, "lr": 6.765388394756504e-05, "epoch": 1, "t": 2.7191402912139893}
|
| 131 |
+
{"step": 131, "loss": 1.192399501800537, "grad_norm": 0.6298495531082153, "lr": 6.715640313457733e-05, "epoch": 1, "t": 3.0565907955169678}
|
| 132 |
+
{"step": 132, "loss": 1.0720741748809814, "grad_norm": 0.5506334900856018, "lr": 6.665698973710288e-05, "epoch": 1, "t": 3.1156535148620605}
|
| 133 |
+
{"step": 133, "loss": 1.1196759939193726, "grad_norm": 0.642974853515625, "lr": 6.615570001160626e-05, "epoch": 1, "t": 2.7423505783081055}
|
| 134 |
+
{"step": 134, "loss": 1.2106388807296753, "grad_norm": 0.5607022047042847, "lr": 6.565259042591113e-05, "epoch": 1, "t": 3.329364061355591}
|
| 135 |
+
{"step": 135, "loss": 1.1919975280761719, "grad_norm": 0.9858223795890808, "lr": 6.514771765283942e-05, "epoch": 1, "t": 3.089970588684082}
|
| 136 |
+
{"step": 136, "loss": 1.1731981039047241, "grad_norm": 0.5917448997497559, "lr": 6.464113856382752e-05, "epoch": 1, "t": 2.911637783050537}
|
| 137 |
+
{"step": 137, "loss": 1.1506966352462769, "grad_norm": 0.5532222986221313, "lr": 6.413291022251989e-05, "epoch": 1, "t": 2.8609025478363037}
|
| 138 |
+
{"step": 138, "loss": 1.2350622415542603, "grad_norm": 0.5943459868431091, "lr": 6.362308987834115e-05, "epoch": 1, "t": 2.9749345779418945}
|
| 139 |
+
{"step": 139, "loss": 1.1088007688522339, "grad_norm": 0.5870715975761414, "lr": 6.311173496004723e-05, "epoch": 1, "t": 2.9706966876983643}
|
| 140 |
+
{"step": 140, "loss": 1.1821991205215454, "grad_norm": 0.6302564144134521, "lr": 6.259890306925627e-05, "epoch": 1, "t": 2.806385040283203}
|
| 141 |
+
{"step": 141, "loss": 1.1470372676849365, "grad_norm": 0.5550669431686401, "lr": 6.208465197396013e-05, "epoch": 1, "t": 2.8569416999816895}
|
| 142 |
+
{"step": 142, "loss": 1.1341872215270996, "grad_norm": 0.6414320468902588, "lr": 6.156903960201709e-05, "epoch": 1, "t": 3.155444383621216}
|
| 143 |
+
{"step": 143, "loss": 1.1604533195495605, "grad_norm": 0.5938991904258728, "lr": 6.105212403462651e-05, "epoch": 1, "t": 3.0608088970184326}
|
| 144 |
+
{"step": 144, "loss": 1.1787281036376953, "grad_norm": 0.6795621514320374, "lr": 6.0533963499786314e-05, "epoch": 1, "t": 2.4089560508728027}
|
| 145 |
+
{"step": 145, "loss": 1.1644045114517212, "grad_norm": 0.563539981842041, "lr": 6.001461636573397e-05, "epoch": 1, "t": 2.9390783309936523}
|
| 146 |
+
{"step": 146, "loss": 1.1728127002716064, "grad_norm": 0.5525833964347839, "lr": 5.949414113437142e-05, "epoch": 1, "t": 3.100515127182007}
|
| 147 |
+
{"step": 147, "loss": 1.1451499462127686, "grad_norm": 0.5328254699707031, "lr": 5.897259643467527e-05, "epoch": 1, "t": 3.0328478813171387}
|
| 148 |
+
{"step": 148, "loss": 1.0825276374816895, "grad_norm": 0.5490713119506836, "lr": 5.8450041016092464e-05, "epoch": 1, "t": 3.075953722000122}
|
| 149 |
+
{"step": 149, "loss": 1.0465219020843506, "grad_norm": 0.7212356328964233, "lr": 5.792653374192245e-05, "epoch": 1, "t": 3.1931333541870117}
|
| 150 |
+
{"step": 150, "loss": 1.1744071245193481, "grad_norm": 0.5588793754577637, "lr": 5.7402133582686576e-05, "epoch": 1, "t": 3.780517101287842}
|
| 151 |
+
{"step": 151, "loss": 1.2614938020706177, "grad_norm": 0.6507968902587891, "lr": 5.6876899609485256e-05, "epoch": 1, "t": 3.3625833988189697}
|
| 152 |
+
{"step": 152, "loss": 1.1556527614593506, "grad_norm": 0.4932103157043457, "lr": 5.6350890987343944e-05, "epoch": 1, "t": 3.547351837158203}
|
| 153 |
+
{"step": 153, "loss": 1.2477785348892212, "grad_norm": 0.5489038228988647, "lr": 5.582416696854853e-05, "epoch": 1, "t": 3.252431631088257}
|
| 154 |
+
{"step": 154, "loss": 1.0835554599761963, "grad_norm": 0.5881918668746948, "lr": 5.5296786885970805e-05, "epoch": 1, "t": 2.5963423252105713}
|
| 155 |
+
{"step": 155, "loss": 1.148956298828125, "grad_norm": 0.5351289510726929, "lr": 5.476881014638491e-05, "epoch": 1, "t": 3.2630696296691895}
|
| 156 |
+
{"step": 156, "loss": 1.2198426723480225, "grad_norm": 0.5861051082611084, "lr": 5.4240296223775465e-05, "epoch": 1, "t": 3.125530958175659}
|
| 157 |
+
{"step": 157, "loss": 1.2282218933105469, "grad_norm": 0.608242928981781, "lr": 5.3711304652638126e-05, "epoch": 1, "t": 3.0210516452789307}
|
| 158 |
+
{"step": 158, "loss": 1.1079331636428833, "grad_norm": 0.6681380271911621, "lr": 5.318189502127332e-05, "epoch": 1, "t": 2.873147487640381}
|
| 159 |
+
{"step": 159, "loss": 1.214823603630066, "grad_norm": 0.5586912631988525, "lr": 5.265212696507387e-05, "epoch": 1, "t": 2.9733569622039795}
|
| 160 |
+
{"step": 160, "loss": 1.2848204374313354, "grad_norm": 0.5761910676956177, "lr": 5.212206015980742e-05, "epoch": 1, "t": 3.1285524368286133}
|
| 161 |
+
{"step": 161, "loss": 1.1382474899291992, "grad_norm": 0.600112795829773, "lr": 5.159175431489424e-05, "epoch": 1, "t": 2.7901957035064697}
|
| 162 |
+
{"step": 162, "loss": 1.2085801362991333, "grad_norm": 0.5726537704467773, "lr": 5.1061269166681183e-05, "epoch": 1, "t": 3.456279754638672}
|
| 163 |
+
{"step": 163, "loss": 1.0004818439483643, "grad_norm": 0.5713548064231873, "lr": 5.053066447171282e-05, "epoch": 1, "t": 3.11616849899292}
|
| 164 |
+
{"step": 164, "loss": 1.1956971883773804, "grad_norm": 0.5816550850868225, "lr": 5e-05, "epoch": 1, "t": 2.889674663543701}
|
| 165 |
+
{"step": 165, "loss": 1.1377452611923218, "grad_norm": 0.5768589973449707, "lr": 4.94693355282872e-05, "epoch": 1, "t": 3.074821949005127}
|
| 166 |
+
{"step": 166, "loss": 1.154218077659607, "grad_norm": 0.564263105392456, "lr": 4.893873083331882e-05, "epoch": 1, "t": 3.1782240867614746}
|
| 167 |
+
{"step": 167, "loss": 1.16450035572052, "grad_norm": 0.5475992560386658, "lr": 4.840824568510579e-05, "epoch": 1, "t": 3.2882606983184814}
|
| 168 |
+
{"step": 168, "loss": 1.108408808708191, "grad_norm": 0.5658562183380127, "lr": 4.78779398401926e-05, "epoch": 1, "t": 2.895862579345703}
|
| 169 |
+
{"step": 169, "loss": 1.2103993892669678, "grad_norm": 0.5811935663223267, "lr": 4.734787303492615e-05, "epoch": 1, "t": 3.062185525894165}
|
| 170 |
+
{"step": 170, "loss": 1.2354953289031982, "grad_norm": 0.6475352048873901, "lr": 4.6818104978726685e-05, "epoch": 1, "t": 3.1329429149627686}
|
| 171 |
+
{"step": 171, "loss": 1.2203470468521118, "grad_norm": 0.5723701119422913, "lr": 4.628869534736187e-05, "epoch": 1, "t": 3.3884599208831787}
|
| 172 |
+
{"step": 172, "loss": 1.1049220561981201, "grad_norm": 0.6683981418609619, "lr": 4.575970377622456e-05, "epoch": 1, "t": 2.925100803375244}
|
| 173 |
+
{"step": 173, "loss": 1.1134202480316162, "grad_norm": 0.5766351222991943, "lr": 4.52311898536151e-05, "epoch": 1, "t": 2.9758660793304443}
|
| 174 |
+
{"step": 174, "loss": 1.151648998260498, "grad_norm": 0.569556713104248, "lr": 4.47032131140292e-05, "epoch": 1, "t": 2.924151659011841}
|
| 175 |
+
{"step": 175, "loss": 1.2023348808288574, "grad_norm": 0.5935404896736145, "lr": 4.4175833031451473e-05, "epoch": 1, "t": 3.134366989135742}
|
| 176 |
+
{"step": 176, "loss": 1.121922254562378, "grad_norm": 0.5166159868240356, "lr": 4.364910901265606e-05, "epoch": 1, "t": 4.2262864112854}
|
| 177 |
+
{"step": 177, "loss": 1.0275864601135254, "grad_norm": 0.5652241110801697, "lr": 4.3123100390514756e-05, "epoch": 1, "t": 4.112625360488892}
|
| 178 |
+
{"step": 178, "loss": 1.1473710536956787, "grad_norm": 0.5857097506523132, "lr": 4.2597866417313436e-05, "epoch": 1, "t": 3.582775831222534}
|
| 179 |
+
{"step": 179, "loss": 1.1719067096710205, "grad_norm": 0.6241664290428162, "lr": 4.207346625807756e-05, "epoch": 1, "t": 3.819958209991455}
|
| 180 |
+
{"step": 180, "loss": 1.1645557880401611, "grad_norm": 0.5976747870445251, "lr": 4.1549958983907555e-05, "epoch": 1, "t": 3.8587818145751953}
|
| 181 |
+
{"step": 181, "loss": 1.1692571640014648, "grad_norm": 0.6483241319656372, "lr": 4.102740356532473e-05, "epoch": 1, "t": 3.7793633937835693}
|
| 182 |
+
{"step": 182, "loss": 1.1669251918792725, "grad_norm": 0.721358597278595, "lr": 4.050585886562858e-05, "epoch": 1, "t": 3.937937021255493}
|
| 183 |
+
{"step": 183, "loss": 1.2466169595718384, "grad_norm": 0.5831030607223511, "lr": 3.998538363426605e-05, "epoch": 1, "t": 3.9405360221862793}
|
| 184 |
+
{"step": 184, "loss": 1.2401959896087646, "grad_norm": 0.5450506806373596, "lr": 3.94660365002137e-05, "epoch": 1, "t": 3.6224429607391357}
|
| 185 |
+
{"step": 185, "loss": 1.209920048713684, "grad_norm": 0.5808328986167908, "lr": 3.894787596537352e-05, "epoch": 1, "t": 4.1509692668914795}
|
| 186 |
+
{"step": 186, "loss": 1.222599744796753, "grad_norm": 0.5833186507225037, "lr": 3.843096039798293e-05, "epoch": 1, "t": 3.2172698974609375}
|
| 187 |
+
{"step": 187, "loss": 1.262049674987793, "grad_norm": 0.5987293124198914, "lr": 3.791534802603988e-05, "epoch": 1, "t": 3.109898805618286}
|
| 188 |
+
{"step": 188, "loss": 1.0296735763549805, "grad_norm": 0.6398347616195679, "lr": 3.740109693074375e-05, "epoch": 1, "t": 2.5187463760375977}
|
| 189 |
+
{"step": 189, "loss": 1.2416787147521973, "grad_norm": 0.528292715549469, "lr": 3.68882650399528e-05, "epoch": 1, "t": 3.6282477378845215}
|
| 190 |
+
{"step": 190, "loss": 1.237912893295288, "grad_norm": 0.6176980137825012, "lr": 3.637691012165886e-05, "epoch": 1, "t": 2.839493751525879}
|
| 191 |
+
{"step": 191, "loss": 1.018993854522705, "grad_norm": 0.6205204129219055, "lr": 3.586708977748012e-05, "epoch": 1, "t": 2.6434435844421387}
|
| 192 |
+
{"step": 192, "loss": 1.125335693359375, "grad_norm": 0.5940316319465637, "lr": 3.5358861436172485e-05, "epoch": 1, "t": 3.1232852935791016}
|
| 193 |
+
{"step": 193, "loss": 1.1238465309143066, "grad_norm": 0.5678236484527588, "lr": 3.485228234716058e-05, "epoch": 1, "t": 3.0870141983032227}
|
| 194 |
+
{"step": 194, "loss": 1.1776199340820312, "grad_norm": 0.5769335031509399, "lr": 3.434740957408889e-05, "epoch": 1, "t": 3.1048550605773926}
|
| 195 |
+
{"step": 195, "loss": 1.0105640888214111, "grad_norm": 0.6299868822097778, "lr": 3.3844299988393755e-05, "epoch": 1, "t": 2.6997573375701904}
|
| 196 |
+
{"step": 196, "loss": 1.2175965309143066, "grad_norm": 0.5081778168678284, "lr": 3.334301026289712e-05, "epoch": 1, "t": 3.4862453937530518}
|
| 197 |
+
{"step": 197, "loss": 1.2957203388214111, "grad_norm": 0.6115744709968567, "lr": 3.284359686542269e-05, "epoch": 1, "t": 2.991338014602661}
|
| 198 |
+
{"step": 198, "loss": 1.0697333812713623, "grad_norm": 0.7884970307350159, "lr": 3.234611605243496e-05, "epoch": 1, "t": 2.6753690242767334}
|
| 199 |
+
{"step": 199, "loss": 1.111758828163147, "grad_norm": 0.5366118550300598, "lr": 3.1850623862702344e-05, "epoch": 1, "t": 3.0785419940948486}
|
| 200 |
+
{"step": 200, "loss": 1.1966639757156372, "grad_norm": 0.5983626246452332, "lr": 3.135717611098458e-05, "epoch": 1, "t": 3.112166404724121}
|
| 201 |
+
{"step": 201, "loss": 1.1379112005233765, "grad_norm": 0.7048618197441101, "lr": 3.086582838174551e-05, "epoch": 1, "t": 3.0757322311401367}
|
| 202 |
+
{"step": 202, "loss": 1.1819733381271362, "grad_norm": 0.5848366618156433, "lr": 3.0376636022891812e-05, "epoch": 1, "t": 2.906740665435791}
|
| 203 |
+
{"step": 203, "loss": 1.0744045972824097, "grad_norm": 0.6132264137268066, "lr": 2.9889654139538246e-05, "epoch": 1, "t": 2.871115207672119}
|
| 204 |
+
{"step": 204, "loss": 1.1134111881256104, "grad_norm": 0.6606585383415222, "lr": 2.9404937587800375e-05, "epoch": 1, "t": 3.229006052017212}
|
| 205 |
+
{"step": 205, "loss": 1.160618782043457, "grad_norm": 0.6119295954704285, "lr": 2.8922540968615286e-05, "epoch": 1, "t": 2.674818277359009}
|
| 206 |
+
{"step": 206, "loss": 1.1156063079833984, "grad_norm": 0.5940641164779663, "lr": 2.8442518621591086e-05, "epoch": 1, "t": 3.035911798477173}
|
| 207 |
+
{"step": 207, "loss": 1.1710171699523926, "grad_norm": 0.6113779544830322, "lr": 2.7964924618885778e-05, "epoch": 1, "t": 3.090020179748535}
|
| 208 |
+
{"step": 208, "loss": 1.160750150680542, "grad_norm": 0.5737333297729492, "lr": 2.748981275911633e-05, "epoch": 1, "t": 3.3775105476379395}
|
| 209 |
+
{"step": 209, "loss": 1.1280392408370972, "grad_norm": 0.6196173429489136, "lr": 2.701723656129851e-05, "epoch": 1, "t": 2.7091281414031982}
|
| 210 |
+
{"step": 210, "loss": 1.1538275480270386, "grad_norm": 0.5874179005622864, "lr": 2.6547249258818164e-05, "epoch": 1, "t": 2.960157632827759}
|
| 211 |
+
{"step": 211, "loss": 1.0798423290252686, "grad_norm": 0.6555476188659668, "lr": 2.607990379343489e-05, "epoch": 1, "t": 3.0391311645507812}
|
| 212 |
+
{"step": 212, "loss": 1.0290166139602661, "grad_norm": 0.6177952885627747, "lr": 2.5615252809318284e-05, "epoch": 1, "t": 2.7866854667663574}
|
| 213 |
+
{"step": 213, "loss": 1.276257038116455, "grad_norm": 0.6012812852859497, "lr": 2.5153348647117857e-05, "epoch": 1, "t": 3.388942241668701}
|
| 214 |
+
{"step": 214, "loss": 1.1391267776489258, "grad_norm": 0.5409966111183167, "lr": 2.469424333806718e-05, "epoch": 1, "t": 3.1688072681427}
|
| 215 |
+
{"step": 215, "loss": 1.036207675933838, "grad_norm": 0.5911509990692139, "lr": 2.4237988598122752e-05, "epoch": 1, "t": 2.798635482788086}
|
| 216 |
+
{"step": 216, "loss": 1.2101538181304932, "grad_norm": 0.6286398768424988, "lr": 2.3784635822138424e-05, "epoch": 1, "t": 2.8621058464050293}
|
| 217 |
+
{"step": 217, "loss": 1.1943871974945068, "grad_norm": 0.5897751450538635, "lr": 2.333423607807613e-05, "epoch": 1, "t": 3.2273385524749756}
|
| 218 |
+
{"step": 218, "loss": 1.1797982454299927, "grad_norm": 0.5689539909362793, "lr": 2.288684010125325e-05, "epoch": 1, "t": 3.2338011264801025}
|
| 219 |
+
{"step": 219, "loss": 1.1917088031768799, "grad_norm": 0.5907241106033325, "lr": 2.2442498288627556e-05, "epoch": 1, "t": 2.9131131172180176}
|
| 220 |
+
{"step": 220, "loss": 1.0297861099243164, "grad_norm": 0.6247298717498779, "lr": 2.2001260693120233e-05, "epoch": 1, "t": 3.043391466140747}
|
| 221 |
+
{"step": 221, "loss": 1.0811527967453003, "grad_norm": 0.6107983589172363, "lr": 2.156317701797766e-05, "epoch": 1, "t": 2.947765350341797}
|
| 222 |
+
{"step": 222, "loss": 1.1478102207183838, "grad_norm": 1.2381247282028198, "lr": 2.1128296611172593e-05, "epoch": 1, "t": 3.040334463119507}
|
| 223 |
+
{"step": 223, "loss": 1.0906646251678467, "grad_norm": 0.633804440498352, "lr": 2.0696668459845355e-05, "epoch": 1, "t": 2.8082356452941895}
|
| 224 |
+
{"step": 224, "loss": 1.2812585830688477, "grad_norm": 0.5863341689109802, "lr": 2.026834118478567e-05, "epoch": 1, "t": 3.1412811279296875}
|
| 225 |
+
{"step": 225, "loss": 1.1530014276504517, "grad_norm": 0.6071820259094238, "lr": 1.98433630349558e-05, "epoch": 1, "t": 3.0048422813415527}
|
| 226 |
+
{"step": 226, "loss": 1.0425373315811157, "grad_norm": 0.607322633266449, "lr": 1.9421781882055444e-05, "epoch": 1, "t": 2.575441360473633}
|
| 227 |
+
{"step": 227, "loss": 1.0592705011367798, "grad_norm": 0.6060351133346558, "lr": 1.9003645215129355e-05, "epoch": 1, "t": 2.9081881046295166}
|
| 228 |
+
{"step": 228, "loss": 1.3345788717269897, "grad_norm": 0.6063174605369568, "lr": 1.858900013521788e-05, "epoch": 1, "t": 3.3664608001708984}
|
| 229 |
+
{"step": 229, "loss": 1.1068217754364014, "grad_norm": 0.5254011154174805, "lr": 1.817789335005121e-05, "epoch": 1, "t": 3.466855525970459}
|
| 230 |
+
{"step": 230, "loss": 1.2544689178466797, "grad_norm": 0.5797206163406372, "lr": 1.777037116878804e-05, "epoch": 1, "t": 3.313175678253174}
|
| 231 |
+
{"step": 231, "loss": 1.2279372215270996, "grad_norm": 0.6262298226356506, "lr": 1.7366479496799077e-05, "epoch": 1, "t": 3.0753684043884277}
|
| 232 |
+
{"step": 232, "loss": 1.1724395751953125, "grad_norm": 0.6728183031082153, "lr": 1.6966263830495936e-05, "epoch": 1, "t": 3.021501064300537}
|
| 233 |
+
{"step": 233, "loss": 1.1790351867675781, "grad_norm": 0.5918564200401306, "lr": 1.656976925220633e-05, "epoch": 1, "t": 3.1134839057922363}
|
| 234 |
+
{"step": 234, "loss": 1.1428561210632324, "grad_norm": 0.5992364287376404, "lr": 1.6177040425095662e-05, "epoch": 1, "t": 3.3391077518463135}
|
| 235 |
+
{"step": 235, "loss": 1.2134824991226196, "grad_norm": 0.6032678484916687, "lr": 1.5788121588135975e-05, "epoch": 1, "t": 3.229142427444458}
|
| 236 |
+
{"step": 236, "loss": 1.0925897359848022, "grad_norm": 0.5905115008354187, "lr": 1.5403056551122697e-05, "epoch": 1, "t": 3.118264675140381}
|
| 237 |
+
{"step": 237, "loss": 1.1609774827957153, "grad_norm": 0.526172935962677, "lr": 1.5021888689739549e-05, "epoch": 1, "t": 3.1487526893615723}
|
| 238 |
+
{"step": 238, "loss": 1.1890206336975098, "grad_norm": 0.597624659538269, "lr": 1.4644660940672627e-05, "epoch": 1, "t": 2.9986464977264404}
|
| 239 |
+
{"step": 239, "loss": 1.162978172302246, "grad_norm": 0.5727379322052002, "lr": 1.427141579677374e-05, "epoch": 1, "t": 2.8487298488616943}
|
| 240 |
+
{"step": 240, "loss": 1.1339913606643677, "grad_norm": 0.559514582157135, "lr": 1.3902195302273779e-05, "epoch": 1, "t": 2.9655044078826904}
|
| 241 |
+
{"step": 241, "loss": 1.1060378551483154, "grad_norm": 0.7390354871749878, "lr": 1.3537041048046695e-05, "epoch": 1, "t": 3.3367416858673096}
|
| 242 |
+
{"step": 242, "loss": 1.1456468105316162, "grad_norm": 0.6592360734939575, "lr": 1.3175994166924394e-05, "epoch": 1, "t": 2.8595519065856934}
|
| 243 |
+
{"step": 243, "loss": 1.0963239669799805, "grad_norm": 0.6013798713684082, "lr": 1.2819095329063469e-05, "epoch": 1, "t": 2.780035972595215}
|
| 244 |
+
{"step": 244, "loss": 1.095710039138794, "grad_norm": 0.6500201225280762, "lr": 1.246638473736378e-05, "epoch": 1, "t": 2.949089527130127}
|
| 245 |
+
{"step": 245, "loss": 1.1575127840042114, "grad_norm": 0.5583919882774353, "lr": 1.2117902122939861e-05, "epoch": 1, "t": 3.335636854171753}
|
| 246 |
+
{"step": 246, "loss": 1.2061023712158203, "grad_norm": 0.8218300938606262, "lr": 1.1773686740645384e-05, "epoch": 1, "t": 3.421555519104004}
|
| 247 |
+
{"step": 247, "loss": 1.0667352676391602, "grad_norm": 0.6150872707366943, "lr": 1.1433777364651271e-05, "epoch": 1, "t": 2.8088629245758057}
|
| 248 |
+
{"step": 248, "loss": 0.9942640066146851, "grad_norm": 0.6881025433540344, "lr": 1.1098212284078036e-05, "epoch": 1, "t": 2.8369979858398438}
|
| 249 |
+
{"step": 249, "loss": 1.1674474477767944, "grad_norm": 0.6190013885498047, "lr": 1.076702929868264e-05, "epoch": 1, "t": 3.0134074687957764}
|
| 250 |
+
{"step": 250, "loss": 1.0568673610687256, "grad_norm": 0.5610866546630859, "lr": 1.0440265714600572e-05, "epoch": 1, "t": 3.058701276779175}
|
| 251 |
+
{"step": 251, "loss": 1.1656662225723267, "grad_norm": 0.626024603843689, "lr": 1.0117958340143507e-05, "epoch": 1, "t": 3.2128448486328125}
|
| 252 |
+
{"step": 252, "loss": 1.1809203624725342, "grad_norm": 0.5629134178161621, "lr": 9.800143481652979e-06, "epoch": 1, "t": 3.2645459175109863}
|
| 253 |
+
{"step": 253, "loss": 1.2196834087371826, "grad_norm": 0.6032998561859131, "lr": 9.48685693941067e-06, "epoch": 1, "t": 2.9141271114349365}
|
| 254 |
+
{"step": 254, "loss": 1.1928837299346924, "grad_norm": 0.6172772645950317, "lr": 9.17813400360572e-06, "epoch": 1, "t": 3.172384262084961}
|
| 255 |
+
{"step": 255, "loss": 1.1708760261535645, "grad_norm": 0.8942421674728394, "lr": 8.874009450359427e-06, "epoch": 1, "t": 2.9025299549102783}
|
| 256 |
+
{"step": 256, "loss": 1.0920674800872803, "grad_norm": 0.574086606502533, "lr": 8.574517537807897e-06, "epoch": 1, "t": 3.108200788497925}
|
| 257 |
+
{"step": 257, "loss": 1.156648874282837, "grad_norm": 0.7499485611915588, "lr": 8.279692002243027e-06, "epoch": 1, "t": 2.6387264728546143}
|
| 258 |
+
{"step": 258, "loss": 1.0950714349746704, "grad_norm": 0.5602185726165771, "lr": 7.989566054312287e-06, "epoch": 1, "t": 3.169079542160034}
|
| 259 |
+
{"step": 259, "loss": 1.0397440195083618, "grad_norm": 0.6062813997268677, "lr": 7.704172375277691e-06, "epoch": 1, "t": 2.9886317253112793}
|
| 260 |
+
{"step": 260, "loss": 1.1563323736190796, "grad_norm": 0.5651493072509766, "lr": 7.423543113334436e-06, "epoch": 1, "t": 2.990959644317627}
|
| 261 |
+
{"step": 261, "loss": 1.1134047508239746, "grad_norm": 0.6180574297904968, "lr": 7.14770987998954e-06, "epoch": 1, "t": 2.8323421478271484}
|
| 262 |
+
{"step": 262, "loss": 1.1044129133224487, "grad_norm": 0.5693318247795105, "lr": 6.876703746500984e-06, "epoch": 1, "t": 3.243398427963257}
|
| 263 |
+
{"step": 263, "loss": 1.1611313819885254, "grad_norm": 0.6057265400886536, "lr": 6.610555240377652e-06, "epoch": 1, "t": 3.2190983295440674}
|
| 264 |
+
{"step": 264, "loss": 1.165337085723877, "grad_norm": 1.1039265394210815, "lr": 6.349294341940593e-06, "epoch": 1, "t": 3.071765422821045}
|
| 265 |
+
{"step": 265, "loss": 1.234431266784668, "grad_norm": 0.6330640316009521, "lr": 6.092950480945897e-06, "epoch": 1, "t": 3.0204012393951416}
|
| 266 |
+
{"step": 266, "loss": 1.127256155014038, "grad_norm": 0.5789159536361694, "lr": 5.841552533269534e-06, "epoch": 1, "t": 3.1319258213043213}
|
| 267 |
+
{"step": 267, "loss": 1.2207940816879272, "grad_norm": 0.5734661221504211, "lr": 5.595128817654638e-06, "epoch": 1, "t": 2.931506633758545}
|
| 268 |
+
{"step": 268, "loss": 1.141287922859192, "grad_norm": 0.5549564957618713, "lr": 5.353707092521582e-06, "epoch": 1, "t": 3.2252449989318848}
|
| 269 |
+
{"step": 269, "loss": 1.125415325164795, "grad_norm": 0.6042464971542358, "lr": 5.117314552841052e-06, "epoch": 1, "t": 2.8536031246185303}
|
| 270 |
+
{"step": 270, "loss": 1.152937412261963, "grad_norm": 0.6050695776939392, "lr": 4.885977827070748e-06, "epoch": 1, "t": 2.954955816268921}
|
| 271 |
+
{"step": 271, "loss": 1.1014562845230103, "grad_norm": 0.572132408618927, "lr": 4.659722974155767e-06, "epoch": 1, "t": 2.971269130706787}
|
| 272 |
+
{"step": 272, "loss": 1.204144835472107, "grad_norm": 0.6065317392349243, "lr": 4.43857548059321e-06, "epoch": 1, "t": 3.043729782104492}
|
| 273 |
+
{"step": 273, "loss": 1.1075233221054077, "grad_norm": 0.5877376198768616, "lr": 4.2225602575612755e-06, "epoch": 1, "t": 3.121161699295044}
|
| 274 |
+
{"step": 274, "loss": 1.2027723789215088, "grad_norm": 0.6455724835395813, "lr": 4.011701638113063e-06, "epoch": 1, "t": 2.8101646900177}
|
| 275 |
+
{"step": 275, "loss": 1.0392215251922607, "grad_norm": 1.4909194707870483, "lr": 3.8060233744356633e-06, "epoch": 1, "t": 3.1290743350982666}
|
| 276 |
+
{"step": 276, "loss": 1.1897180080413818, "grad_norm": 0.7497091293334961, "lr": 3.605548635174533e-06, "epoch": 1, "t": 2.9680233001708984}
|
| 277 |
+
{"step": 277, "loss": 1.1153748035430908, "grad_norm": 0.565839946269989, "lr": 3.410300002823691e-06, "epoch": 1, "t": 2.9037387371063232}
|
| 278 |
+
{"step": 278, "loss": 1.1494334936141968, "grad_norm": 0.5831167101860046, "lr": 3.220299471181898e-06, "epoch": 1, "t": 3.2570207118988037}
|
| 279 |
+
{"step": 279, "loss": 1.222648024559021, "grad_norm": 0.5716479420661926, "lr": 3.035568442875136e-06, "epoch": 1, "t": 2.9446775913238525}
|
| 280 |
+
{"step": 280, "loss": 1.2059191465377808, "grad_norm": 0.5954435467720032, "lr": 2.85612772694579e-06, "epoch": 1, "t": 3.4033713340759277}
|
| 281 |
+
{"step": 281, "loss": 1.1299675703048706, "grad_norm": 0.595647931098938, "lr": 2.6819975365085237e-06, "epoch": 1, "t": 2.9016120433807373}
|
| 282 |
+
{"step": 282, "loss": 1.167212963104248, "grad_norm": 0.5784906148910522, "lr": 2.5131974864734066e-06, "epoch": 1, "t": 3.159802198410034}
|
| 283 |
+
{"step": 283, "loss": 1.1120587587356567, "grad_norm": 0.546314537525177, "lr": 2.349746591336405e-06, "epoch": 1, "t": 3.116912364959717}
|
| 284 |
+
{"step": 284, "loss": 1.2408440113067627, "grad_norm": 0.5839245915412903, "lr": 2.191663263037458e-06, "epoch": 1, "t": 3.47688889503479}
|
| 285 |
+
{"step": 285, "loss": 1.044086217880249, "grad_norm": 0.5622262358665466, "lr": 2.0389653088865036e-06, "epoch": 1, "t": 3.3095884323120117}
|
| 286 |
+
{"step": 286, "loss": 1.1614607572555542, "grad_norm": 0.6686910390853882, "lr": 1.8916699295575324e-06, "epoch": 1, "t": 3.385456085205078}
|
| 287 |
+
{"step": 287, "loss": 1.1427057981491089, "grad_norm": 0.6589815616607666, "lr": 1.7497937171510547e-06, "epoch": 1, "t": 3.5058515071868896}
|
| 288 |
+
{"step": 288, "loss": 1.1672608852386475, "grad_norm": 0.5413684248924255, "lr": 1.6133526533250565e-06, "epoch": 1, "t": 3.5682971477508545}
|
| 289 |
+
{"step": 289, "loss": 1.2121870517730713, "grad_norm": 0.579685389995575, "lr": 1.4823621074947503e-06, "epoch": 1, "t": 3.607837677001953}
|
| 290 |
+
{"step": 290, "loss": 1.2169768810272217, "grad_norm": 0.7050402760505676, "lr": 1.3568368351012717e-06, "epoch": 1, "t": 3.1011247634887695}
|
| 291 |
+
{"step": 291, "loss": 0.9908221960067749, "grad_norm": 0.5955489873886108, "lr": 1.236790975949592e-06, "epoch": 1, "t": 2.782078504562378}
|
| 292 |
+
{"step": 292, "loss": 1.181696891784668, "grad_norm": 0.6093128323554993, "lr": 1.1222380526156928e-06, "epoch": 1, "t": 3.1663551330566406}
|
| 293 |
+
{"step": 293, "loss": 1.0274240970611572, "grad_norm": 0.5747653245925903, "lr": 1.0131909689233442e-06, "epoch": 1, "t": 2.8262951374053955}
|
| 294 |
+
{"step": 294, "loss": 1.1584563255310059, "grad_norm": 0.5272875428199768, "lr": 9.096620084905472e-07, "epoch": 1, "t": 3.1543571949005127}
|
| 295 |
+
{"step": 295, "loss": 1.161163091659546, "grad_norm": 0.9364246129989624, "lr": 8.11662833345822e-07, "epoch": 1, "t": 3.2010724544525146}
|
| 296 |
+
{"step": 296, "loss": 1.2536511421203613, "grad_norm": 0.7115570306777954, "lr": 7.192044826145771e-07, "epoch": 1, "t": 2.9636471271514893}
|
| 297 |
+
{"step": 297, "loss": 1.08464515209198, "grad_norm": 0.6501760482788086, "lr": 6.322973712755697e-07, "epoch": 1, "t": 3.114466667175293}
|
| 298 |
+
{"step": 298, "loss": 1.173872470855713, "grad_norm": 0.5717540979385376, "lr": 5.509512889877333e-07, "epoch": 1, "t": 3.058483839035034}
|
| 299 |
+
{"step": 299, "loss": 1.142005205154419, "grad_norm": 0.6201208829879761, "lr": 4.7517539898741524e-07, "epoch": 1, "t": 2.8155181407928467}
|
| 300 |
+
{"step": 300, "loss": 1.084489345550537, "grad_norm": 0.5723140239715576, "lr": 4.049782370561583e-07, "epoch": 1, "t": 2.8282744884490967}
|
| 301 |
+
{"step": 301, "loss": 1.194441795349121, "grad_norm": 0.6134630441665649, "lr": 3.4036771055923066e-07, "epoch": 1, "t": 3.1520586013793945}
|
| 302 |
+
{"step": 302, "loss": 1.2328448295593262, "grad_norm": 0.6585222482681274, "lr": 2.813510975548772e-07, "epoch": 1, "t": 3.0300822257995605}
|
| 303 |
+
{"step": 303, "loss": 1.0778056383132935, "grad_norm": 0.6462317109107971, "lr": 2.2793504597447002e-07, "epoch": 1, "t": 2.920275926589966}
|
| 304 |
+
{"step": 304, "loss": 1.1771395206451416, "grad_norm": 0.6489781141281128, "lr": 1.8012557287367392e-07, "epoch": 1, "t": 2.778437614440918}
|
| 305 |
+
{"step": 305, "loss": 1.2051841020584106, "grad_norm": 0.6162313222885132, "lr": 1.379280637546443e-07, "epoch": 1, "t": 3.040475845336914}
|
| 306 |
+
{"step": 306, "loss": 1.1684272289276123, "grad_norm": 0.6035457253456116, "lr": 1.0134727195937333e-07, "epoch": 1, "t": 3.256772041320801}
|
| 307 |
+
{"step": 307, "loss": 1.1923434734344482, "grad_norm": 0.6025996804237366, "lr": 7.038731813426291e-08, "epoch": 1, "t": 2.972303628921509}
|
| 308 |
+
{"step": 308, "loss": 1.2231829166412354, "grad_norm": 0.6525549292564392, "lr": 4.5051689765929214e-08, "epoch": 1, "t": 3.1314585208892822}
|
| 309 |
+
{"step": 309, "loss": 1.0794585943222046, "grad_norm": 0.6763843297958374, "lr": 2.534324078837802e-08, "epoch": 1, "t": 3.2646920680999756}
|
| 310 |
+
{"step": 310, "loss": 1.1793197393417358, "grad_norm": 0.5490835905075073, "lr": 1.1264191261528557e-08, "epoch": 1, "t": 3.2566683292388916}
|
| 311 |
+
{"step": 311, "loss": 1.071898341178894, "grad_norm": 0.6597471833229065, "lr": 2.8161271211024633e-09, "epoch": 1, "t": 2.7107973098754883}
|
| 312 |
+
{"step": 312, "loss": 1.12252676486969, "grad_norm": 0.5467647910118103, "lr": 0.0, "epoch": 1, "t": 3.132575273513794}
|
12b/V_top-s0/README.md
ADDED
|
@@ -0,0 +1,206 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: google/gemma-3-12b-it
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- base_model:adapter:google/gemma-3-12b-it
|
| 6 |
+
- lora
|
| 7 |
+
- transformers
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# Model Card for Model ID
|
| 11 |
+
|
| 12 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
## Model Details
|
| 17 |
+
|
| 18 |
+
### Model Description
|
| 19 |
+
|
| 20 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
- **Developed by:** [More Information Needed]
|
| 25 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 26 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 27 |
+
- **Model type:** [More Information Needed]
|
| 28 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 29 |
+
- **License:** [More Information Needed]
|
| 30 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 31 |
+
|
| 32 |
+
### Model Sources [optional]
|
| 33 |
+
|
| 34 |
+
<!-- Provide the basic links for the model. -->
|
| 35 |
+
|
| 36 |
+
- **Repository:** [More Information Needed]
|
| 37 |
+
- **Paper [optional]:** [More Information Needed]
|
| 38 |
+
- **Demo [optional]:** [More Information Needed]
|
| 39 |
+
|
| 40 |
+
## Uses
|
| 41 |
+
|
| 42 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 43 |
+
|
| 44 |
+
### Direct Use
|
| 45 |
+
|
| 46 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 47 |
+
|
| 48 |
+
[More Information Needed]
|
| 49 |
+
|
| 50 |
+
### Downstream Use [optional]
|
| 51 |
+
|
| 52 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 53 |
+
|
| 54 |
+
[More Information Needed]
|
| 55 |
+
|
| 56 |
+
### Out-of-Scope Use
|
| 57 |
+
|
| 58 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 59 |
+
|
| 60 |
+
[More Information Needed]
|
| 61 |
+
|
| 62 |
+
## Bias, Risks, and Limitations
|
| 63 |
+
|
| 64 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 65 |
+
|
| 66 |
+
[More Information Needed]
|
| 67 |
+
|
| 68 |
+
### Recommendations
|
| 69 |
+
|
| 70 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 71 |
+
|
| 72 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 73 |
+
|
| 74 |
+
## How to Get Started with the Model
|
| 75 |
+
|
| 76 |
+
Use the code below to get started with the model.
|
| 77 |
+
|
| 78 |
+
[More Information Needed]
|
| 79 |
+
|
| 80 |
+
## Training Details
|
| 81 |
+
|
| 82 |
+
### Training Data
|
| 83 |
+
|
| 84 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 85 |
+
|
| 86 |
+
[More Information Needed]
|
| 87 |
+
|
| 88 |
+
### Training Procedure
|
| 89 |
+
|
| 90 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 91 |
+
|
| 92 |
+
#### Preprocessing [optional]
|
| 93 |
+
|
| 94 |
+
[More Information Needed]
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
#### Training Hyperparameters
|
| 98 |
+
|
| 99 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 100 |
+
|
| 101 |
+
#### Speeds, Sizes, Times [optional]
|
| 102 |
+
|
| 103 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 104 |
+
|
| 105 |
+
[More Information Needed]
|
| 106 |
+
|
| 107 |
+
## Evaluation
|
| 108 |
+
|
| 109 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 110 |
+
|
| 111 |
+
### Testing Data, Factors & Metrics
|
| 112 |
+
|
| 113 |
+
#### Testing Data
|
| 114 |
+
|
| 115 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 116 |
+
|
| 117 |
+
[More Information Needed]
|
| 118 |
+
|
| 119 |
+
#### Factors
|
| 120 |
+
|
| 121 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 122 |
+
|
| 123 |
+
[More Information Needed]
|
| 124 |
+
|
| 125 |
+
#### Metrics
|
| 126 |
+
|
| 127 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 128 |
+
|
| 129 |
+
[More Information Needed]
|
| 130 |
+
|
| 131 |
+
### Results
|
| 132 |
+
|
| 133 |
+
[More Information Needed]
|
| 134 |
+
|
| 135 |
+
#### Summary
|
| 136 |
+
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
## Model Examination [optional]
|
| 140 |
+
|
| 141 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 142 |
+
|
| 143 |
+
[More Information Needed]
|
| 144 |
+
|
| 145 |
+
## Environmental Impact
|
| 146 |
+
|
| 147 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 148 |
+
|
| 149 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 150 |
+
|
| 151 |
+
- **Hardware Type:** [More Information Needed]
|
| 152 |
+
- **Hours used:** [More Information Needed]
|
| 153 |
+
- **Cloud Provider:** [More Information Needed]
|
| 154 |
+
- **Compute Region:** [More Information Needed]
|
| 155 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 156 |
+
|
| 157 |
+
## Technical Specifications [optional]
|
| 158 |
+
|
| 159 |
+
### Model Architecture and Objective
|
| 160 |
+
|
| 161 |
+
[More Information Needed]
|
| 162 |
+
|
| 163 |
+
### Compute Infrastructure
|
| 164 |
+
|
| 165 |
+
[More Information Needed]
|
| 166 |
+
|
| 167 |
+
#### Hardware
|
| 168 |
+
|
| 169 |
+
[More Information Needed]
|
| 170 |
+
|
| 171 |
+
#### Software
|
| 172 |
+
|
| 173 |
+
[More Information Needed]
|
| 174 |
+
|
| 175 |
+
## Citation [optional]
|
| 176 |
+
|
| 177 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 178 |
+
|
| 179 |
+
**BibTeX:**
|
| 180 |
+
|
| 181 |
+
[More Information Needed]
|
| 182 |
+
|
| 183 |
+
**APA:**
|
| 184 |
+
|
| 185 |
+
[More Information Needed]
|
| 186 |
+
|
| 187 |
+
## Glossary [optional]
|
| 188 |
+
|
| 189 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 190 |
+
|
| 191 |
+
[More Information Needed]
|
| 192 |
+
|
| 193 |
+
## More Information [optional]
|
| 194 |
+
|
| 195 |
+
[More Information Needed]
|
| 196 |
+
|
| 197 |
+
## Model Card Authors [optional]
|
| 198 |
+
|
| 199 |
+
[More Information Needed]
|
| 200 |
+
|
| 201 |
+
## Model Card Contact
|
| 202 |
+
|
| 203 |
+
[More Information Needed]
|
| 204 |
+
### Framework versions
|
| 205 |
+
|
| 206 |
+
- PEFT 0.17.1
|
12b/V_top-s0/adapter_config.json
ADDED
|
@@ -0,0 +1,186 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": {
|
| 4 |
+
"base_model_class": "Gemma3ForConditionalGeneration",
|
| 5 |
+
"parent_library": "transformers.models.gemma3.modeling_gemma3"
|
| 6 |
+
},
|
| 7 |
+
"base_model_name_or_path": "google/gemma-3-12b-it",
|
| 8 |
+
"bias": "none",
|
| 9 |
+
"corda_config": null,
|
| 10 |
+
"eva_config": null,
|
| 11 |
+
"exclude_modules": null,
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": true,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layer_replication": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"layers_to_transform": null,
|
| 18 |
+
"loftq_config": {},
|
| 19 |
+
"lora_alpha": 128,
|
| 20 |
+
"lora_bias": false,
|
| 21 |
+
"lora_dropout": 0.05,
|
| 22 |
+
"megatron_config": null,
|
| 23 |
+
"megatron_core": "megatron.core",
|
| 24 |
+
"modules_to_save": null,
|
| 25 |
+
"peft_type": "LORA",
|
| 26 |
+
"qalora_group_size": 16,
|
| 27 |
+
"r": 64,
|
| 28 |
+
"rank_pattern": {},
|
| 29 |
+
"revision": null,
|
| 30 |
+
"target_modules": [
|
| 31 |
+
"47.self_attn.k_proj",
|
| 32 |
+
"language_model.layers.7.self_attn.k_proj",
|
| 33 |
+
"43.self_attn.q_proj",
|
| 34 |
+
"language_model.layers.4.self_attn.q_proj",
|
| 35 |
+
"language_model.layers.18.self_attn.k_proj",
|
| 36 |
+
"language_model.layers.25.self_attn.v_proj",
|
| 37 |
+
"27.self_attn.q_proj",
|
| 38 |
+
"language_model.layers.8.self_attn.k_proj",
|
| 39 |
+
"language_model.layers.14.self_attn.k_proj",
|
| 40 |
+
"language_model.layers.20.self_attn.k_proj",
|
| 41 |
+
"language_model.layers.20.self_attn.q_proj",
|
| 42 |
+
"language_model.layers.0.self_attn.k_proj",
|
| 43 |
+
"38.self_attn.k_proj",
|
| 44 |
+
"46.self_attn.q_proj",
|
| 45 |
+
"44.self_attn.k_proj",
|
| 46 |
+
"language_model.layers.14.self_attn.v_proj",
|
| 47 |
+
"34.self_attn.k_proj",
|
| 48 |
+
"41.self_attn.k_proj",
|
| 49 |
+
"39.self_attn.q_proj",
|
| 50 |
+
"language_model.layers.16.self_attn.k_proj",
|
| 51 |
+
"31.self_attn.q_proj",
|
| 52 |
+
"41.self_attn.v_proj",
|
| 53 |
+
"language_model.layers.10.self_attn.k_proj",
|
| 54 |
+
"language_model.layers.2.self_attn.q_proj",
|
| 55 |
+
"29.self_attn.k_proj",
|
| 56 |
+
"28.self_attn.k_proj",
|
| 57 |
+
"language_model.layers.8.self_attn.v_proj",
|
| 58 |
+
"language_model.layers.5.self_attn.v_proj",
|
| 59 |
+
"34.self_attn.q_proj",
|
| 60 |
+
"35.self_attn.q_proj",
|
| 61 |
+
"31.self_attn.k_proj",
|
| 62 |
+
"40.self_attn.k_proj",
|
| 63 |
+
"language_model.layers.5.self_attn.k_proj",
|
| 64 |
+
"language_model.layers.10.self_attn.q_proj",
|
| 65 |
+
"36.self_attn.v_proj",
|
| 66 |
+
"language_model.layers.20.self_attn.v_proj",
|
| 67 |
+
"44.self_attn.q_proj",
|
| 68 |
+
"language_model.layers.26.self_attn.k_proj",
|
| 69 |
+
"language_model.layers.9.self_attn.q_proj",
|
| 70 |
+
"language_model.layers.6.self_attn.v_proj",
|
| 71 |
+
"language_model.layers.23.self_attn.k_proj",
|
| 72 |
+
"35.self_attn.k_proj",
|
| 73 |
+
"32.self_attn.q_proj",
|
| 74 |
+
"language_model.layers.3.self_attn.q_proj",
|
| 75 |
+
"42.self_attn.v_proj",
|
| 76 |
+
"language_model.layers.19.self_attn.v_proj",
|
| 77 |
+
"language_model.layers.18.self_attn.q_proj",
|
| 78 |
+
"language_model.layers.16.self_attn.v_proj",
|
| 79 |
+
"43.self_attn.k_proj",
|
| 80 |
+
"44.self_attn.v_proj",
|
| 81 |
+
"language_model.layers.23.self_attn.q_proj",
|
| 82 |
+
"28.self_attn.q_proj",
|
| 83 |
+
"30.self_attn.k_proj",
|
| 84 |
+
"language_model.layers.17.self_attn.v_proj",
|
| 85 |
+
"language_model.layers.0.self_attn.q_proj",
|
| 86 |
+
"language_model.layers.22.self_attn.v_proj",
|
| 87 |
+
"language_model.layers.5.self_attn.q_proj",
|
| 88 |
+
"language_model.layers.22.self_attn.q_proj",
|
| 89 |
+
"language_model.layers.25.self_attn.q_proj",
|
| 90 |
+
"language_model.layers.6.self_attn.q_proj",
|
| 91 |
+
"38.self_attn.v_proj",
|
| 92 |
+
"39.self_attn.k_proj",
|
| 93 |
+
"language_model.layers.14.self_attn.q_proj",
|
| 94 |
+
"language_model.layers.10.self_attn.v_proj",
|
| 95 |
+
"37.self_attn.q_proj",
|
| 96 |
+
"language_model.layers.19.self_attn.k_proj",
|
| 97 |
+
"language_model.layers.18.self_attn.v_proj",
|
| 98 |
+
"45.self_attn.k_proj",
|
| 99 |
+
"42.self_attn.q_proj",
|
| 100 |
+
"38.self_attn.q_proj",
|
| 101 |
+
"30.self_attn.v_proj",
|
| 102 |
+
"o_proj",
|
| 103 |
+
"29.self_attn.q_proj",
|
| 104 |
+
"language_model.layers.17.self_attn.k_proj",
|
| 105 |
+
"language_model.layers.7.self_attn.q_proj",
|
| 106 |
+
"language_model.layers.4.self_attn.k_proj",
|
| 107 |
+
"language_model.layers.24.self_attn.k_proj",
|
| 108 |
+
"39.self_attn.v_proj",
|
| 109 |
+
"46.self_attn.v_proj",
|
| 110 |
+
"language_model.layers.1.self_attn.q_proj",
|
| 111 |
+
"43.self_attn.v_proj",
|
| 112 |
+
"language_model.layers.1.self_attn.v_proj",
|
| 113 |
+
"language_model.layers.12.self_attn.q_proj",
|
| 114 |
+
"33.self_attn.q_proj",
|
| 115 |
+
"46.self_attn.k_proj",
|
| 116 |
+
"language_model.layers.9.self_attn.k_proj",
|
| 117 |
+
"language_model.layers.19.self_attn.q_proj",
|
| 118 |
+
"language_model.layers.2.self_attn.v_proj",
|
| 119 |
+
"language_model.layers.9.self_attn.v_proj",
|
| 120 |
+
"30.self_attn.q_proj",
|
| 121 |
+
"37.self_attn.k_proj",
|
| 122 |
+
"41.self_attn.q_proj",
|
| 123 |
+
"language_model.layers.16.self_attn.q_proj",
|
| 124 |
+
"language_model.layers.22.self_attn.k_proj",
|
| 125 |
+
"language_model.layers.21.self_attn.v_proj",
|
| 126 |
+
"language_model.layers.21.self_attn.q_proj",
|
| 127 |
+
"language_model.layers.23.self_attn.v_proj",
|
| 128 |
+
"36.self_attn.q_proj",
|
| 129 |
+
"33.self_attn.v_proj",
|
| 130 |
+
"language_model.layers.26.self_attn.v_proj",
|
| 131 |
+
"language_model.layers.1.self_attn.k_proj",
|
| 132 |
+
"language_model.layers.12.self_attn.v_proj",
|
| 133 |
+
"37.self_attn.v_proj",
|
| 134 |
+
"33.self_attn.k_proj",
|
| 135 |
+
"language_model.layers.4.self_attn.v_proj",
|
| 136 |
+
"language_model.layers.25.self_attn.k_proj",
|
| 137 |
+
"language_model.layers.0.self_attn.v_proj",
|
| 138 |
+
"language_model.layers.11.self_attn.q_proj",
|
| 139 |
+
"27.self_attn.k_proj",
|
| 140 |
+
"language_model.layers.3.self_attn.v_proj",
|
| 141 |
+
"language_model.layers.8.self_attn.q_proj",
|
| 142 |
+
"36.self_attn.k_proj",
|
| 143 |
+
"language_model.layers.17.self_attn.q_proj",
|
| 144 |
+
"language_model.layers.24.self_attn.q_proj",
|
| 145 |
+
"language_model.layers.6.self_attn.k_proj",
|
| 146 |
+
"32.self_attn.k_proj",
|
| 147 |
+
"down_proj",
|
| 148 |
+
"language_model.layers.24.self_attn.v_proj",
|
| 149 |
+
"language_model.layers.12.self_attn.k_proj",
|
| 150 |
+
"32.self_attn.v_proj",
|
| 151 |
+
"47.self_attn.v_proj",
|
| 152 |
+
"language_model.layers.13.self_attn.k_proj",
|
| 153 |
+
"40.self_attn.v_proj",
|
| 154 |
+
"language_model.layers.11.self_attn.k_proj",
|
| 155 |
+
"language_model.layers.13.self_attn.v_proj",
|
| 156 |
+
"27.self_attn.v_proj",
|
| 157 |
+
"47.self_attn.q_proj",
|
| 158 |
+
"45.self_attn.q_proj",
|
| 159 |
+
"29.self_attn.v_proj",
|
| 160 |
+
"42.self_attn.k_proj",
|
| 161 |
+
"35.self_attn.v_proj",
|
| 162 |
+
"language_model.layers.2.self_attn.k_proj",
|
| 163 |
+
"40.self_attn.q_proj",
|
| 164 |
+
"28.self_attn.v_proj",
|
| 165 |
+
"34.self_attn.v_proj",
|
| 166 |
+
"31.self_attn.v_proj",
|
| 167 |
+
"language_model.layers.15.self_attn.k_proj",
|
| 168 |
+
"gate_proj",
|
| 169 |
+
"language_model.layers.13.self_attn.q_proj",
|
| 170 |
+
"language_model.layers.11.self_attn.v_proj",
|
| 171 |
+
"language_model.layers.21.self_attn.k_proj",
|
| 172 |
+
"45.self_attn.v_proj",
|
| 173 |
+
"language_model.layers.15.self_attn.q_proj",
|
| 174 |
+
"language_model.layers.26.self_attn.q_proj",
|
| 175 |
+
"up_proj",
|
| 176 |
+
"language_model.layers.7.self_attn.v_proj",
|
| 177 |
+
"language_model.layers.15.self_attn.v_proj",
|
| 178 |
+
"language_model.layers.3.self_attn.k_proj"
|
| 179 |
+
],
|
| 180 |
+
"target_parameters": null,
|
| 181 |
+
"task_type": null,
|
| 182 |
+
"trainable_token_indices": null,
|
| 183 |
+
"use_dora": false,
|
| 184 |
+
"use_qalora": false,
|
| 185 |
+
"use_rslora": false
|
| 186 |
+
}
|
12b/V_top-s0/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ada804409e6ea599986b3e44fb8d8acfe61c4992ff292104f8bc548ab0891add
|
| 3 |
+
size 1047628488
|
12b/V_top-s0/config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "google/gemma-3-12b-it",
|
| 3 |
+
"data": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma12/arms_v/TASKW-V_top-gemma12.jsonl",
|
| 4 |
+
"run": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma12/runs/V_top-s0",
|
| 5 |
+
"steps": 0,
|
| 6 |
+
"batch": 16,
|
| 7 |
+
"micro": 1,
|
| 8 |
+
"seq_len": 4096,
|
| 9 |
+
"lr": 0.0001,
|
| 10 |
+
"warmup": 10,
|
| 11 |
+
"lora_rank": 64,
|
| 12 |
+
"lora_targets": "all-linear",
|
| 13 |
+
"lora_targets_file": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/code/gemma3_12b_lora_targets.json",
|
| 14 |
+
"seed": 0,
|
| 15 |
+
"bos": true,
|
| 16 |
+
"no_merge": false,
|
| 17 |
+
"save_at": "",
|
| 18 |
+
"replay": "",
|
| 19 |
+
"replay_frac": 0.0,
|
| 20 |
+
"constant_lr": false,
|
| 21 |
+
"schedule": "hf-cosine",
|
| 22 |
+
"min_lr_frac": 0.0,
|
| 23 |
+
"warmup_ratio": 0.05,
|
| 24 |
+
"lora_dropout": 0.05,
|
| 25 |
+
"no_pack": true,
|
| 26 |
+
"chat_sft": true,
|
| 27 |
+
"data_frac": 1.0,
|
| 28 |
+
"base_adapter": "",
|
| 29 |
+
"ckpt_every": 0,
|
| 30 |
+
"kill_after": 0,
|
| 31 |
+
"method": "doc_lm_lora",
|
| 32 |
+
"world": 2
|
| 33 |
+
}
|
12b/V_top-s0/log.jsonl
ADDED
|
@@ -0,0 +1,312 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 1, "loss": 2.2729265689849854, "grad_norm": 4.406494617462158, "lr": 1.25e-05, "epoch": 1, "t": 3.4905970096588135}
|
| 2 |
+
{"step": 2, "loss": 2.141500473022461, "grad_norm": 3.7048614025115967, "lr": 1.8750000000000002e-05, "epoch": 1, "t": 3.6400585174560547}
|
| 3 |
+
{"step": 3, "loss": 2.157907485961914, "grad_norm": 3.6539409160614014, "lr": 2.5e-05, "epoch": 1, "t": 3.1290602684020996}
|
| 4 |
+
{"step": 4, "loss": 1.9312562942504883, "grad_norm": 2.9819629192352295, "lr": 3.125e-05, "epoch": 1, "t": 2.7889459133148193}
|
| 5 |
+
{"step": 5, "loss": 1.8254902362823486, "grad_norm": 2.225151300430298, "lr": 3.7500000000000003e-05, "epoch": 1, "t": 3.802821397781372}
|
| 6 |
+
{"step": 6, "loss": 1.7398107051849365, "grad_norm": 1.4814915657043457, "lr": 4.375e-05, "epoch": 1, "t": 3.2151715755462646}
|
| 7 |
+
{"step": 7, "loss": 1.723631739616394, "grad_norm": 1.0053844451904297, "lr": 5e-05, "epoch": 1, "t": 3.005499839782715}
|
| 8 |
+
{"step": 8, "loss": 1.5290982723236084, "grad_norm": 0.8869999647140503, "lr": 5.6250000000000005e-05, "epoch": 1, "t": 3.2517714500427246}
|
| 9 |
+
{"step": 9, "loss": 1.715596318244934, "grad_norm": 1.0236982107162476, "lr": 6.25e-05, "epoch": 1, "t": 3.0355679988861084}
|
| 10 |
+
{"step": 10, "loss": 1.6154892444610596, "grad_norm": 1.0504403114318848, "lr": 6.875e-05, "epoch": 1, "t": 3.108259916305542}
|
| 11 |
+
{"step": 11, "loss": 1.6134092807769775, "grad_norm": 0.990784764289856, "lr": 7.500000000000001e-05, "epoch": 1, "t": 2.8776910305023193}
|
| 12 |
+
{"step": 12, "loss": 1.3646092414855957, "grad_norm": 0.8541951775550842, "lr": 8.125000000000001e-05, "epoch": 1, "t": 3.1841750144958496}
|
| 13 |
+
{"step": 13, "loss": 1.2742877006530762, "grad_norm": 0.769559919834137, "lr": 8.75e-05, "epoch": 1, "t": 3.058722496032715}
|
| 14 |
+
{"step": 14, "loss": 1.4351048469543457, "grad_norm": 0.7996724247932434, "lr": 9.375e-05, "epoch": 1, "t": 3.130711793899536}
|
| 15 |
+
{"step": 15, "loss": 1.51334547996521, "grad_norm": 0.7810581922531128, "lr": 0.0001, "epoch": 1, "t": 3.1117939949035645}
|
| 16 |
+
{"step": 16, "loss": 1.4208080768585205, "grad_norm": 0.7467895150184631, "lr": 0.0001, "epoch": 1, "t": 3.283149242401123}
|
| 17 |
+
{"step": 17, "loss": 1.3618955612182617, "grad_norm": 0.6926639080047607, "lr": 9.99971838728789e-05, "epoch": 1, "t": 3.1131327152252197}
|
| 18 |
+
{"step": 18, "loss": 1.3402419090270996, "grad_norm": 0.734306812286377, "lr": 9.998873580873848e-05, "epoch": 1, "t": 3.040325403213501}
|
| 19 |
+
{"step": 19, "loss": 1.362644076347351, "grad_norm": 0.7452945709228516, "lr": 9.997465675921163e-05, "epoch": 1, "t": 3.3454794883728027}
|
| 20 |
+
{"step": 20, "loss": 1.3237106800079346, "grad_norm": 0.622259259223938, "lr": 9.995494831023409e-05, "epoch": 1, "t": 3.157949447631836}
|
| 21 |
+
{"step": 21, "loss": 1.219679832458496, "grad_norm": 0.6535102128982544, "lr": 9.992961268186573e-05, "epoch": 1, "t": 3.2318668365478516}
|
| 22 |
+
{"step": 22, "loss": 1.3675858974456787, "grad_norm": 0.8337335586547852, "lr": 9.989865272804063e-05, "epoch": 1, "t": 3.0391271114349365}
|
| 23 |
+
{"step": 23, "loss": 1.1954965591430664, "grad_norm": 0.6240617036819458, "lr": 9.986207193624536e-05, "epoch": 1, "t": 3.4128594398498535}
|
| 24 |
+
{"step": 24, "loss": 1.3337948322296143, "grad_norm": 0.5964548587799072, "lr": 9.981987442712633e-05, "epoch": 1, "t": 3.1717514991760254}
|
| 25 |
+
{"step": 25, "loss": 1.131766676902771, "grad_norm": 0.587676465511322, "lr": 9.977206495402554e-05, "epoch": 1, "t": 3.2898590564727783}
|
| 26 |
+
{"step": 26, "loss": 1.3129777908325195, "grad_norm": 0.6612595319747925, "lr": 9.971864890244513e-05, "epoch": 1, "t": 3.2893996238708496}
|
| 27 |
+
{"step": 27, "loss": 1.3074558973312378, "grad_norm": 0.6082188487052917, "lr": 9.965963228944078e-05, "epoch": 1, "t": 3.0682950019836426}
|
| 28 |
+
{"step": 28, "loss": 1.2593557834625244, "grad_norm": 0.5356127023696899, "lr": 9.959502176294383e-05, "epoch": 1, "t": 3.553260326385498}
|
| 29 |
+
{"step": 29, "loss": 1.2174677848815918, "grad_norm": 0.7074805498123169, "lr": 9.95248246010126e-05, "epoch": 1, "t": 3.072585344314575}
|
| 30 |
+
{"step": 30, "loss": 1.3831980228424072, "grad_norm": 0.5847600698471069, "lr": 9.944904871101228e-05, "epoch": 1, "t": 3.1686363220214844}
|
| 31 |
+
{"step": 31, "loss": 1.3148130178451538, "grad_norm": 0.649976909160614, "lr": 9.936770262872443e-05, "epoch": 1, "t": 2.7638027667999268}
|
| 32 |
+
{"step": 32, "loss": 1.3472368717193604, "grad_norm": 0.6375584602355957, "lr": 9.928079551738543e-05, "epoch": 1, "t": 3.3834211826324463}
|
| 33 |
+
{"step": 33, "loss": 1.2677112817764282, "grad_norm": 0.5937733054161072, "lr": 9.918833716665419e-05, "epoch": 1, "t": 3.0703229904174805}
|
| 34 |
+
{"step": 34, "loss": 1.2804527282714844, "grad_norm": 0.9339198470115662, "lr": 9.909033799150946e-05, "epoch": 1, "t": 3.434422016143799}
|
| 35 |
+
{"step": 35, "loss": 1.3502745628356934, "grad_norm": 0.713164746761322, "lr": 9.898680903107666e-05, "epoch": 1, "t": 3.1491732597351074}
|
| 36 |
+
{"step": 36, "loss": 1.2597880363464355, "grad_norm": 0.5872796773910522, "lr": 9.887776194738432e-05, "epoch": 1, "t": 3.5551655292510986}
|
| 37 |
+
{"step": 37, "loss": 1.255307912826538, "grad_norm": 0.6113024353981018, "lr": 9.876320902405042e-05, "epoch": 1, "t": 3.5403289794921875}
|
| 38 |
+
{"step": 38, "loss": 1.0736026763916016, "grad_norm": 0.5962145328521729, "lr": 9.864316316489873e-05, "epoch": 1, "t": 2.9595353603363037}
|
| 39 |
+
{"step": 39, "loss": 1.1489514112472534, "grad_norm": 0.6147931218147278, "lr": 9.851763789250525e-05, "epoch": 1, "t": 3.266500473022461}
|
| 40 |
+
{"step": 40, "loss": 1.2019245624542236, "grad_norm": 0.64427649974823, "lr": 9.838664734667495e-05, "epoch": 1, "t": 3.3068037033081055}
|
| 41 |
+
{"step": 41, "loss": 1.2238175868988037, "grad_norm": 0.5384893417358398, "lr": 9.825020628284896e-05, "epoch": 1, "t": 2.8172926902770996}
|
| 42 |
+
{"step": 42, "loss": 1.3091044425964355, "grad_norm": 0.6122305989265442, "lr": 9.810833007044247e-05, "epoch": 1, "t": 3.186645269393921}
|
| 43 |
+
{"step": 43, "loss": 1.323056936264038, "grad_norm": 0.6110641956329346, "lr": 9.796103469111351e-05, "epoch": 1, "t": 3.412797212600708}
|
| 44 |
+
{"step": 44, "loss": 1.1668975353240967, "grad_norm": 0.573988139629364, "lr": 9.780833673696254e-05, "epoch": 1, "t": 3.426820993423462}
|
| 45 |
+
{"step": 45, "loss": 1.270620346069336, "grad_norm": 0.5737598538398743, "lr": 9.76502534086636e-05, "epoch": 1, "t": 3.084648609161377}
|
| 46 |
+
{"step": 46, "loss": 1.2267508506774902, "grad_norm": 0.6040119528770447, "lr": 9.74868025135266e-05, "epoch": 1, "t": 3.2545011043548584}
|
| 47 |
+
{"step": 47, "loss": 1.2665650844573975, "grad_norm": 0.6519939303398132, "lr": 9.731800246349148e-05, "epoch": 1, "t": 3.2667903900146484}
|
| 48 |
+
{"step": 48, "loss": 1.22287917137146, "grad_norm": 0.6127481460571289, "lr": 9.714387227305422e-05, "epoch": 1, "t": 3.124504804611206}
|
| 49 |
+
{"step": 49, "loss": 1.1991313695907593, "grad_norm": 0.6090063452720642, "lr": 9.696443155712486e-05, "epoch": 1, "t": 3.0702877044677734}
|
| 50 |
+
{"step": 50, "loss": 1.2632099390029907, "grad_norm": 0.6437141299247742, "lr": 9.67797005288181e-05, "epoch": 1, "t": 3.05598783493042}
|
| 51 |
+
{"step": 51, "loss": 1.344907522201538, "grad_norm": 0.5727119445800781, "lr": 9.65896999971763e-05, "epoch": 1, "t": 3.2655839920043945}
|
| 52 |
+
{"step": 52, "loss": 1.1435203552246094, "grad_norm": 0.5720223188400269, "lr": 9.639445136482548e-05, "epoch": 1, "t": 3.4344396591186523}
|
| 53 |
+
{"step": 53, "loss": 1.2046306133270264, "grad_norm": 0.7183040380477905, "lr": 9.619397662556435e-05, "epoch": 1, "t": 3.1110618114471436}
|
| 54 |
+
{"step": 54, "loss": 1.3263490200042725, "grad_norm": 0.641219973564148, "lr": 9.598829836188694e-05, "epoch": 1, "t": 3.1065995693206787}
|
| 55 |
+
{"step": 55, "loss": 1.3485932350158691, "grad_norm": 0.6010202169418335, "lr": 9.577743974243874e-05, "epoch": 1, "t": 3.5663695335388184}
|
| 56 |
+
{"step": 56, "loss": 1.1829782724380493, "grad_norm": 0.5923372507095337, "lr": 9.55614245194068e-05, "epoch": 1, "t": 3.588824987411499}
|
| 57 |
+
{"step": 57, "loss": 1.265611171722412, "grad_norm": 0.6543408632278442, "lr": 9.534027702584425e-05, "epoch": 1, "t": 3.52703595161438}
|
| 58 |
+
{"step": 58, "loss": 1.1645268201828003, "grad_norm": 0.6018239259719849, "lr": 9.511402217292926e-05, "epoch": 1, "t": 3.0196752548217773}
|
| 59 |
+
{"step": 59, "loss": 1.2357110977172852, "grad_norm": 0.6383073925971985, "lr": 9.488268544715896e-05, "epoch": 1, "t": 3.346628427505493}
|
| 60 |
+
{"step": 60, "loss": 1.1333954334259033, "grad_norm": 0.5893785953521729, "lr": 9.464629290747842e-05, "epoch": 1, "t": 3.4075937271118164}
|
| 61 |
+
{"step": 61, "loss": 1.371953010559082, "grad_norm": 0.6793717741966248, "lr": 9.440487118234535e-05, "epoch": 1, "t": 3.11741304397583}
|
| 62 |
+
{"step": 62, "loss": 1.270416259765625, "grad_norm": 0.6013386845588684, "lr": 9.415844746673047e-05, "epoch": 1, "t": 2.8872106075286865}
|
| 63 |
+
{"step": 63, "loss": 1.158923864364624, "grad_norm": 0.5537129640579224, "lr": 9.390704951905411e-05, "epoch": 1, "t": 3.417269706726074}
|
| 64 |
+
{"step": 64, "loss": 1.2678682804107666, "grad_norm": 0.6210922598838806, "lr": 9.365070565805941e-05, "epoch": 1, "t": 3.2121496200561523}
|
| 65 |
+
{"step": 65, "loss": 1.2033791542053223, "grad_norm": 0.6727984547615051, "lr": 9.338944475962237e-05, "epoch": 1, "t": 3.0121853351593018}
|
| 66 |
+
{"step": 66, "loss": 1.1171982288360596, "grad_norm": 0.6295392513275146, "lr": 9.312329625349902e-05, "epoch": 1, "t": 3.172537326812744}
|
| 67 |
+
{"step": 67, "loss": 1.31270432472229, "grad_norm": 0.6367724537849426, "lr": 9.285229012001047e-05, "epoch": 1, "t": 3.2105793952941895}
|
| 68 |
+
{"step": 68, "loss": 1.222433090209961, "grad_norm": 0.5490050911903381, "lr": 9.257645688666556e-05, "epoch": 1, "t": 3.302910566329956}
|
| 69 |
+
{"step": 69, "loss": 1.2453744411468506, "grad_norm": 0.6330830454826355, "lr": 9.22958276247223e-05, "epoch": 1, "t": 2.7399702072143555}
|
| 70 |
+
{"step": 70, "loss": 1.2824645042419434, "grad_norm": 0.6556730270385742, "lr": 9.201043394568773e-05, "epoch": 1, "t": 2.9631824493408203}
|
| 71 |
+
{"step": 71, "loss": 1.0487356185913086, "grad_norm": 0.6062486171722412, "lr": 9.172030799775699e-05, "epoch": 1, "t": 3.266761541366577}
|
| 72 |
+
{"step": 72, "loss": 1.2581334114074707, "grad_norm": 0.5626318454742432, "lr": 9.142548246219212e-05, "epoch": 1, "t": 2.8180768489837646}
|
| 73 |
+
{"step": 73, "loss": 1.0210614204406738, "grad_norm": 0.7015877962112427, "lr": 9.112599054964057e-05, "epoch": 1, "t": 3.0019402503967285}
|
| 74 |
+
{"step": 74, "loss": 1.1685079336166382, "grad_norm": 0.6257519721984863, "lr": 9.082186599639428e-05, "epoch": 1, "t": 3.162327527999878}
|
| 75 |
+
{"step": 75, "loss": 1.0390830039978027, "grad_norm": 0.57488614320755, "lr": 9.051314306058933e-05, "epoch": 1, "t": 2.8893725872039795}
|
| 76 |
+
{"step": 76, "loss": 1.1523704528808594, "grad_norm": 0.6371343731880188, "lr": 9.019985651834703e-05, "epoch": 1, "t": 3.662313938140869}
|
| 77 |
+
{"step": 77, "loss": 1.2368154525756836, "grad_norm": 0.6149131059646606, "lr": 8.988204165985649e-05, "epoch": 1, "t": 3.5036165714263916}
|
| 78 |
+
{"step": 78, "loss": 1.2111961841583252, "grad_norm": 0.6960669755935669, "lr": 8.955973428539944e-05, "epoch": 1, "t": 3.2394280433654785}
|
| 79 |
+
{"step": 79, "loss": 1.2572574615478516, "grad_norm": 0.5682839751243591, "lr": 8.923297070131737e-05, "epoch": 1, "t": 3.3606321811676025}
|
| 80 |
+
{"step": 80, "loss": 1.222373127937317, "grad_norm": 0.6081323623657227, "lr": 8.890178771592199e-05, "epoch": 1, "t": 3.0673413276672363}
|
| 81 |
+
{"step": 81, "loss": 1.2898247241973877, "grad_norm": 0.6645525097846985, "lr": 8.856622263534875e-05, "epoch": 1, "t": 3.016792058944702}
|
| 82 |
+
{"step": 82, "loss": 1.3272919654846191, "grad_norm": 0.6964982151985168, "lr": 8.822631325935463e-05, "epoch": 1, "t": 2.723292350769043}
|
| 83 |
+
{"step": 83, "loss": 1.3005414009094238, "grad_norm": 0.6550720930099487, "lr": 8.788209787706015e-05, "epoch": 1, "t": 3.2887024879455566}
|
| 84 |
+
{"step": 84, "loss": 1.2012135982513428, "grad_norm": 0.7282535433769226, "lr": 8.753361526263621e-05, "epoch": 1, "t": 3.040131092071533}
|
| 85 |
+
{"step": 85, "loss": 1.0954171419143677, "grad_norm": 0.7641703486442566, "lr": 8.718090467093654e-05, "epoch": 1, "t": 2.938715696334839}
|
| 86 |
+
{"step": 86, "loss": 1.0850834846496582, "grad_norm": 0.5393131971359253, "lr": 8.682400583307562e-05, "epoch": 1, "t": 3.2568469047546387}
|
| 87 |
+
{"step": 87, "loss": 1.2838066816329956, "grad_norm": 0.5829893350601196, "lr": 8.646295895195333e-05, "epoch": 1, "t": 3.154959201812744}
|
| 88 |
+
{"step": 88, "loss": 1.0138061046600342, "grad_norm": 1.0342228412628174, "lr": 8.609780469772623e-05, "epoch": 1, "t": 3.298295736312866}
|
| 89 |
+
{"step": 89, "loss": 1.1860487461090088, "grad_norm": 0.6082722544670105, "lr": 8.572858420322627e-05, "epoch": 1, "t": 3.3159265518188477}
|
| 90 |
+
{"step": 90, "loss": 1.2038296461105347, "grad_norm": 0.5841742753982544, "lr": 8.535533905932738e-05, "epoch": 1, "t": 3.3225772380828857}
|
| 91 |
+
{"step": 91, "loss": 1.206883430480957, "grad_norm": 0.6440202593803406, "lr": 8.497811131026046e-05, "epoch": 1, "t": 3.4104104042053223}
|
| 92 |
+
{"step": 92, "loss": 1.197068214416504, "grad_norm": 0.5538234710693359, "lr": 8.459694344887732e-05, "epoch": 1, "t": 3.493060350418091}
|
| 93 |
+
{"step": 93, "loss": 1.1775360107421875, "grad_norm": 0.6358944177627563, "lr": 8.421187841186402e-05, "epoch": 1, "t": 2.902132272720337}
|
| 94 |
+
{"step": 94, "loss": 1.28639817237854, "grad_norm": 0.6550676822662354, "lr": 8.382295957490436e-05, "epoch": 1, "t": 3.133793354034424}
|
| 95 |
+
{"step": 95, "loss": 1.2666467428207397, "grad_norm": 0.5645681619644165, "lr": 8.343023074779368e-05, "epoch": 1, "t": 3.51326847076416}
|
| 96 |
+
{"step": 96, "loss": 1.209169626235962, "grad_norm": 0.5733826160430908, "lr": 8.303373616950408e-05, "epoch": 1, "t": 2.9776999950408936}
|
| 97 |
+
{"step": 97, "loss": 1.2243549823760986, "grad_norm": 0.6480118632316589, "lr": 8.263352050320094e-05, "epoch": 1, "t": 3.220109701156616}
|
| 98 |
+
{"step": 98, "loss": 1.1669116020202637, "grad_norm": 0.6220917701721191, "lr": 8.222962883121196e-05, "epoch": 1, "t": 3.0820975303649902}
|
| 99 |
+
{"step": 99, "loss": 1.2751058340072632, "grad_norm": 0.6334115862846375, "lr": 8.182210664994878e-05, "epoch": 1, "t": 3.0170717239379883}
|
| 100 |
+
{"step": 100, "loss": 1.2395230531692505, "grad_norm": 0.6259497404098511, "lr": 8.141099986478212e-05, "epoch": 1, "t": 2.9425978660583496}
|
| 101 |
+
{"step": 101, "loss": 1.1824058294296265, "grad_norm": 0.6447683572769165, "lr": 8.099635478487064e-05, "epoch": 1, "t": 3.1402673721313477}
|
| 102 |
+
{"step": 102, "loss": 1.1586616039276123, "grad_norm": 0.5753451585769653, "lr": 8.057821811794458e-05, "epoch": 1, "t": 3.3552639484405518}
|
| 103 |
+
{"step": 103, "loss": 1.3114122152328491, "grad_norm": 0.6271833181381226, "lr": 8.015663696504422e-05, "epoch": 1, "t": 3.071321725845337}
|
| 104 |
+
{"step": 104, "loss": 1.190492868423462, "grad_norm": 0.6563113331794739, "lr": 7.973165881521434e-05, "epoch": 1, "t": 3.491790771484375}
|
| 105 |
+
{"step": 105, "loss": 1.1693925857543945, "grad_norm": 0.6276815533638, "lr": 7.930333154015466e-05, "epoch": 1, "t": 3.0832626819610596}
|
| 106 |
+
{"step": 106, "loss": 1.2199524641036987, "grad_norm": 0.5738036036491394, "lr": 7.88717033888274e-05, "epoch": 1, "t": 2.834317684173584}
|
| 107 |
+
{"step": 107, "loss": 1.1463149785995483, "grad_norm": 0.584001898765564, "lr": 7.843682298202235e-05, "epoch": 1, "t": 3.0211265087127686}
|
| 108 |
+
{"step": 108, "loss": 1.2491018772125244, "grad_norm": 0.5822736620903015, "lr": 7.799873930687978e-05, "epoch": 1, "t": 3.284855604171753}
|
| 109 |
+
{"step": 109, "loss": 1.2575188875198364, "grad_norm": 0.5760317444801331, "lr": 7.755750171137246e-05, "epoch": 1, "t": 3.3838648796081543}
|
| 110 |
+
{"step": 110, "loss": 1.2893245220184326, "grad_norm": 0.658383309841156, "lr": 7.711315989874677e-05, "epoch": 1, "t": 2.9641647338867188}
|
| 111 |
+
{"step": 111, "loss": 1.2616381645202637, "grad_norm": 0.6391677260398865, "lr": 7.666576392192389e-05, "epoch": 1, "t": 2.9753646850585938}
|
| 112 |
+
{"step": 112, "loss": 1.1679496765136719, "grad_norm": 0.7346526980400085, "lr": 7.621536417786159e-05, "epoch": 1, "t": 3.2575337886810303}
|
| 113 |
+
{"step": 113, "loss": 1.2629032135009766, "grad_norm": 0.6406223177909851, "lr": 7.576201140187727e-05, "epoch": 1, "t": 2.884500503540039}
|
| 114 |
+
{"step": 114, "loss": 1.130014419555664, "grad_norm": 0.5820654630661011, "lr": 7.530575666193283e-05, "epoch": 1, "t": 3.504206657409668}
|
| 115 |
+
{"step": 115, "loss": 1.1340734958648682, "grad_norm": 0.6243785619735718, "lr": 7.484665135288213e-05, "epoch": 1, "t": 3.1025187969207764}
|
| 116 |
+
{"step": 116, "loss": 0.9770622253417969, "grad_norm": 0.6665069460868835, "lr": 7.438474719068173e-05, "epoch": 1, "t": 3.2083888053894043}
|
| 117 |
+
{"step": 117, "loss": 1.1668474674224854, "grad_norm": 0.7202078104019165, "lr": 7.392009620656513e-05, "epoch": 1, "t": 2.960710287094116}
|
| 118 |
+
{"step": 118, "loss": 1.1972293853759766, "grad_norm": 0.670929491519928, "lr": 7.345275074118185e-05, "epoch": 1, "t": 3.0364036560058594}
|
| 119 |
+
{"step": 119, "loss": 1.1133782863616943, "grad_norm": 0.5603964924812317, "lr": 7.298276343870151e-05, "epoch": 1, "t": 3.302926778793335}
|
| 120 |
+
{"step": 120, "loss": 1.1300263404846191, "grad_norm": 0.6096655130386353, "lr": 7.251018724088367e-05, "epoch": 1, "t": 2.788339614868164}
|
| 121 |
+
{"step": 121, "loss": 1.108581304550171, "grad_norm": 0.6814972758293152, "lr": 7.203507538111423e-05, "epoch": 1, "t": 3.142627239227295}
|
| 122 |
+
{"step": 122, "loss": 1.1961642503738403, "grad_norm": 0.6451613903045654, "lr": 7.155748137840892e-05, "epoch": 1, "t": 3.2714130878448486}
|
| 123 |
+
{"step": 123, "loss": 1.2988810539245605, "grad_norm": 0.5895416140556335, "lr": 7.107745903138472e-05, "epoch": 1, "t": 3.262310266494751}
|
| 124 |
+
{"step": 124, "loss": 1.250213384628296, "grad_norm": 0.5825182199478149, "lr": 7.059506241219965e-05, "epoch": 1, "t": 3.3047902584075928}
|
| 125 |
+
{"step": 125, "loss": 1.1177399158477783, "grad_norm": 0.5993546843528748, "lr": 7.011034586046176e-05, "epoch": 1, "t": 3.3100192546844482}
|
| 126 |
+
{"step": 126, "loss": 1.1853973865509033, "grad_norm": 0.5644398927688599, "lr": 6.962336397710819e-05, "epoch": 1, "t": 3.297424554824829}
|
| 127 |
+
{"step": 127, "loss": 1.083348035812378, "grad_norm": 0.6715015172958374, "lr": 6.91341716182545e-05, "epoch": 1, "t": 3.0592894554138184}
|
| 128 |
+
{"step": 128, "loss": 1.1226119995117188, "grad_norm": 0.6135982275009155, "lr": 6.864282388901544e-05, "epoch": 1, "t": 3.140148162841797}
|
| 129 |
+
{"step": 129, "loss": 1.1056246757507324, "grad_norm": 0.6817781329154968, "lr": 6.814937613729766e-05, "epoch": 1, "t": 3.015061855316162}
|
| 130 |
+
{"step": 130, "loss": 1.1483442783355713, "grad_norm": 0.637622594833374, "lr": 6.765388394756504e-05, "epoch": 1, "t": 2.763184070587158}
|
| 131 |
+
{"step": 131, "loss": 1.1246089935302734, "grad_norm": 0.6347431540489197, "lr": 6.715640313457733e-05, "epoch": 1, "t": 3.202000141143799}
|
| 132 |
+
{"step": 132, "loss": 1.0986381769180298, "grad_norm": 0.5413315892219543, "lr": 6.665698973710288e-05, "epoch": 1, "t": 3.249988317489624}
|
| 133 |
+
{"step": 133, "loss": 1.097859501838684, "grad_norm": 0.6063200831413269, "lr": 6.615570001160626e-05, "epoch": 1, "t": 3.0272021293640137}
|
| 134 |
+
{"step": 134, "loss": 1.1629688739776611, "grad_norm": 0.5601895451545715, "lr": 6.565259042591113e-05, "epoch": 1, "t": 3.263235330581665}
|
| 135 |
+
{"step": 135, "loss": 1.1872186660766602, "grad_norm": 0.7107530832290649, "lr": 6.514771765283942e-05, "epoch": 1, "t": 2.9488368034362793}
|
| 136 |
+
{"step": 136, "loss": 1.1221897602081299, "grad_norm": 0.6101102232933044, "lr": 6.464113856382752e-05, "epoch": 1, "t": 3.2652335166931152}
|
| 137 |
+
{"step": 137, "loss": 1.1738827228546143, "grad_norm": 0.5609278678894043, "lr": 6.413291022251989e-05, "epoch": 1, "t": 3.250384569168091}
|
| 138 |
+
{"step": 138, "loss": 1.2568328380584717, "grad_norm": 0.5932942032814026, "lr": 6.362308987834115e-05, "epoch": 1, "t": 3.3005435466766357}
|
| 139 |
+
{"step": 139, "loss": 1.1358098983764648, "grad_norm": 0.6544591784477234, "lr": 6.311173496004723e-05, "epoch": 1, "t": 3.1560237407684326}
|
| 140 |
+
{"step": 140, "loss": 1.194271445274353, "grad_norm": 0.6982852816581726, "lr": 6.259890306925627e-05, "epoch": 1, "t": 3.034360408782959}
|
| 141 |
+
{"step": 141, "loss": 1.1506847143173218, "grad_norm": 0.5750068426132202, "lr": 6.208465197396013e-05, "epoch": 1, "t": 3.097151279449463}
|
| 142 |
+
{"step": 142, "loss": 1.123265027999878, "grad_norm": 0.6349514722824097, "lr": 6.156903960201709e-05, "epoch": 1, "t": 3.2174665927886963}
|
| 143 |
+
{"step": 143, "loss": 1.2196425199508667, "grad_norm": 0.6227642297744751, "lr": 6.105212403462651e-05, "epoch": 1, "t": 3.142275094985962}
|
| 144 |
+
{"step": 144, "loss": 1.1404293775558472, "grad_norm": 0.6641533970832825, "lr": 6.0533963499786314e-05, "epoch": 1, "t": 2.874513864517212}
|
| 145 |
+
{"step": 145, "loss": 1.1745306253433228, "grad_norm": 0.5463564991950989, "lr": 6.001461636573397e-05, "epoch": 1, "t": 3.1237735748291016}
|
| 146 |
+
{"step": 146, "loss": 1.214239239692688, "grad_norm": 0.5621738433837891, "lr": 5.949414113437142e-05, "epoch": 1, "t": 3.3486557006835938}
|
| 147 |
+
{"step": 147, "loss": 1.1681621074676514, "grad_norm": 0.5608762502670288, "lr": 5.897259643467527e-05, "epoch": 1, "t": 3.1366868019104004}
|
| 148 |
+
{"step": 148, "loss": 1.0693703889846802, "grad_norm": 0.557685375213623, "lr": 5.8450041016092464e-05, "epoch": 1, "t": 3.027944803237915}
|
| 149 |
+
{"step": 149, "loss": 1.087806224822998, "grad_norm": 0.6668304800987244, "lr": 5.792653374192245e-05, "epoch": 1, "t": 3.001246690750122}
|
| 150 |
+
{"step": 150, "loss": 1.1747510433197021, "grad_norm": 0.5668718814849854, "lr": 5.7402133582686576e-05, "epoch": 1, "t": 3.1275386810302734}
|
| 151 |
+
{"step": 151, "loss": 1.2637999057769775, "grad_norm": 0.6705304980278015, "lr": 5.6876899609485256e-05, "epoch": 1, "t": 2.971377372741699}
|
| 152 |
+
{"step": 152, "loss": 1.1549617052078247, "grad_norm": 0.49867674708366394, "lr": 5.6350890987343944e-05, "epoch": 1, "t": 3.563704252243042}
|
| 153 |
+
{"step": 153, "loss": 1.2957277297973633, "grad_norm": 0.6017604470252991, "lr": 5.582416696854853e-05, "epoch": 1, "t": 3.4656805992126465}
|
| 154 |
+
{"step": 154, "loss": 1.112256646156311, "grad_norm": 0.6232874393463135, "lr": 5.5296786885970805e-05, "epoch": 1, "t": 3.1391067504882812}
|
| 155 |
+
{"step": 155, "loss": 1.1789627075195312, "grad_norm": 0.5496797561645508, "lr": 5.476881014638491e-05, "epoch": 1, "t": 3.340393543243408}
|
| 156 |
+
{"step": 156, "loss": 1.1763181686401367, "grad_norm": 0.6288057565689087, "lr": 5.4240296223775465e-05, "epoch": 1, "t": 3.0915231704711914}
|
| 157 |
+
{"step": 157, "loss": 1.223020315170288, "grad_norm": 0.646073579788208, "lr": 5.3711304652638126e-05, "epoch": 1, "t": 3.0915229320526123}
|
| 158 |
+
{"step": 158, "loss": 1.1210674047470093, "grad_norm": 0.6813981533050537, "lr": 5.318189502127332e-05, "epoch": 1, "t": 3.022441864013672}
|
| 159 |
+
{"step": 159, "loss": 1.1883270740509033, "grad_norm": 0.5533057451248169, "lr": 5.265212696507387e-05, "epoch": 1, "t": 3.237208366394043}
|
| 160 |
+
{"step": 160, "loss": 1.2698025703430176, "grad_norm": 0.6933168768882751, "lr": 5.212206015980742e-05, "epoch": 1, "t": 3.109149932861328}
|
| 161 |
+
{"step": 161, "loss": 1.1684236526489258, "grad_norm": 0.6078118681907654, "lr": 5.159175431489424e-05, "epoch": 1, "t": 3.052083969116211}
|
| 162 |
+
{"step": 162, "loss": 1.1811952590942383, "grad_norm": 0.5393804311752319, "lr": 5.1061269166681183e-05, "epoch": 1, "t": 3.500868320465088}
|
| 163 |
+
{"step": 163, "loss": 1.050422191619873, "grad_norm": 0.5921630263328552, "lr": 5.053066447171282e-05, "epoch": 1, "t": 3.230645179748535}
|
| 164 |
+
{"step": 164, "loss": 1.2019083499908447, "grad_norm": 0.5943882465362549, "lr": 5e-05, "epoch": 1, "t": 3.409299373626709}
|
| 165 |
+
{"step": 165, "loss": 1.1639859676361084, "grad_norm": 0.5662556886672974, "lr": 4.94693355282872e-05, "epoch": 1, "t": 3.1613926887512207}
|
| 166 |
+
{"step": 166, "loss": 1.1924824714660645, "grad_norm": 0.5759224891662598, "lr": 4.893873083331882e-05, "epoch": 1, "t": 3.4098846912384033}
|
| 167 |
+
{"step": 167, "loss": 1.0863919258117676, "grad_norm": 0.5584909319877625, "lr": 4.840824568510579e-05, "epoch": 1, "t": 3.2176594734191895}
|
| 168 |
+
{"step": 168, "loss": 1.1040669679641724, "grad_norm": 0.590480387210846, "lr": 4.78779398401926e-05, "epoch": 1, "t": 3.0995821952819824}
|
| 169 |
+
{"step": 169, "loss": 1.2440369129180908, "grad_norm": 0.6164286732673645, "lr": 4.734787303492615e-05, "epoch": 1, "t": 3.0424904823303223}
|
| 170 |
+
{"step": 170, "loss": 1.2618274688720703, "grad_norm": 0.5930309891700745, "lr": 4.6818104978726685e-05, "epoch": 1, "t": 3.380537748336792}
|
| 171 |
+
{"step": 171, "loss": 1.257959008216858, "grad_norm": 0.5765829086303711, "lr": 4.628869534736187e-05, "epoch": 1, "t": 3.247805595397949}
|
| 172 |
+
{"step": 172, "loss": 1.0856506824493408, "grad_norm": 0.680612325668335, "lr": 4.575970377622456e-05, "epoch": 1, "t": 3.0945072174072266}
|
| 173 |
+
{"step": 173, "loss": 1.1014269590377808, "grad_norm": 0.6119309067726135, "lr": 4.52311898536151e-05, "epoch": 1, "t": 3.069814920425415}
|
| 174 |
+
{"step": 174, "loss": 1.1343410015106201, "grad_norm": 0.5517014265060425, "lr": 4.47032131140292e-05, "epoch": 1, "t": 3.2381527423858643}
|
| 175 |
+
{"step": 175, "loss": 1.2033703327178955, "grad_norm": 0.6256654858589172, "lr": 4.4175833031451473e-05, "epoch": 1, "t": 2.9093897342681885}
|
| 176 |
+
{"step": 176, "loss": 1.1122409105300903, "grad_norm": 0.5356712937355042, "lr": 4.364910901265606e-05, "epoch": 1, "t": 3.3648529052734375}
|
| 177 |
+
{"step": 177, "loss": 1.0629881620407104, "grad_norm": 0.5948192477226257, "lr": 4.3123100390514756e-05, "epoch": 1, "t": 3.3540995121002197}
|
| 178 |
+
{"step": 178, "loss": 1.0911394357681274, "grad_norm": 0.5890280604362488, "lr": 4.2597866417313436e-05, "epoch": 1, "t": 2.7660906314849854}
|
| 179 |
+
{"step": 179, "loss": 1.1505049467086792, "grad_norm": 0.5762637853622437, "lr": 4.207346625807756e-05, "epoch": 1, "t": 3.26324725151062}
|
| 180 |
+
{"step": 180, "loss": 1.2229969501495361, "grad_norm": 0.6637646555900574, "lr": 4.1549958983907555e-05, "epoch": 1, "t": 3.205742835998535}
|
| 181 |
+
{"step": 181, "loss": 1.1818498373031616, "grad_norm": 0.6205675005912781, "lr": 4.102740356532473e-05, "epoch": 1, "t": 3.0563788414001465}
|
| 182 |
+
{"step": 182, "loss": 1.1463134288787842, "grad_norm": 0.5895227193832397, "lr": 4.050585886562858e-05, "epoch": 1, "t": 3.354653835296631}
|
| 183 |
+
{"step": 183, "loss": 1.2545373439788818, "grad_norm": 0.6045985817909241, "lr": 3.998538363426605e-05, "epoch": 1, "t": 3.2929279804229736}
|
| 184 |
+
{"step": 184, "loss": 1.213509440422058, "grad_norm": 0.5701714754104614, "lr": 3.94660365002137e-05, "epoch": 1, "t": 3.1152875423431396}
|
| 185 |
+
{"step": 185, "loss": 1.1366838216781616, "grad_norm": 0.5886996388435364, "lr": 3.894787596537352e-05, "epoch": 1, "t": 2.9948294162750244}
|
| 186 |
+
{"step": 186, "loss": 1.1989463567733765, "grad_norm": 0.587399423122406, "lr": 3.843096039798293e-05, "epoch": 1, "t": 3.095613718032837}
|
| 187 |
+
{"step": 187, "loss": 1.2452516555786133, "grad_norm": 0.6013587713241577, "lr": 3.791534802603988e-05, "epoch": 1, "t": 3.2710258960723877}
|
| 188 |
+
{"step": 188, "loss": 1.107773780822754, "grad_norm": 0.6943751573562622, "lr": 3.740109693074375e-05, "epoch": 1, "t": 2.88173770904541}
|
| 189 |
+
{"step": 189, "loss": 1.221630334854126, "grad_norm": 0.5519124865531921, "lr": 3.68882650399528e-05, "epoch": 1, "t": 3.6243529319763184}
|
| 190 |
+
{"step": 190, "loss": 1.296262502670288, "grad_norm": 0.6040520071983337, "lr": 3.637691012165886e-05, "epoch": 1, "t": 3.190082550048828}
|
| 191 |
+
{"step": 191, "loss": 1.0229010581970215, "grad_norm": 0.6353663206100464, "lr": 3.586708977748012e-05, "epoch": 1, "t": 3.0634703636169434}
|
| 192 |
+
{"step": 192, "loss": 1.1086941957473755, "grad_norm": 0.5914731025695801, "lr": 3.5358861436172485e-05, "epoch": 1, "t": 2.985412120819092}
|
| 193 |
+
{"step": 193, "loss": 1.152599811553955, "grad_norm": 0.5836315155029297, "lr": 3.485228234716058e-05, "epoch": 1, "t": 3.1702616214752197}
|
| 194 |
+
{"step": 194, "loss": 1.1761579513549805, "grad_norm": 0.6064594388008118, "lr": 3.434740957408889e-05, "epoch": 1, "t": 3.2431304454803467}
|
| 195 |
+
{"step": 195, "loss": 1.0056873559951782, "grad_norm": 0.6169631481170654, "lr": 3.3844299988393755e-05, "epoch": 1, "t": 3.1858839988708496}
|
| 196 |
+
{"step": 196, "loss": 1.188856601715088, "grad_norm": 0.5115482211112976, "lr": 3.334301026289712e-05, "epoch": 1, "t": 3.479349374771118}
|
| 197 |
+
{"step": 197, "loss": 1.26715087890625, "grad_norm": 0.5956758856773376, "lr": 3.284359686542269e-05, "epoch": 1, "t": 3.0514256954193115}
|
| 198 |
+
{"step": 198, "loss": 1.0637043714523315, "grad_norm": 0.6245729327201843, "lr": 3.234611605243496e-05, "epoch": 1, "t": 2.9399797916412354}
|
| 199 |
+
{"step": 199, "loss": 1.1112587451934814, "grad_norm": 0.8566910624504089, "lr": 3.1850623862702344e-05, "epoch": 1, "t": 3.244462728500366}
|
| 200 |
+
{"step": 200, "loss": 1.1548724174499512, "grad_norm": 0.5739861130714417, "lr": 3.135717611098458e-05, "epoch": 1, "t": 3.48917555809021}
|
| 201 |
+
{"step": 201, "loss": 1.1380984783172607, "grad_norm": 0.6413898468017578, "lr": 3.086582838174551e-05, "epoch": 1, "t": 3.262972593307495}
|
| 202 |
+
{"step": 202, "loss": 1.184216856956482, "grad_norm": 0.6022217869758606, "lr": 3.0376636022891812e-05, "epoch": 1, "t": 3.1735434532165527}
|
| 203 |
+
{"step": 203, "loss": 1.0677496194839478, "grad_norm": 0.594731330871582, "lr": 2.9889654139538246e-05, "epoch": 1, "t": 3.0828380584716797}
|
| 204 |
+
{"step": 204, "loss": 1.0658659934997559, "grad_norm": 0.6577484011650085, "lr": 2.9404937587800375e-05, "epoch": 1, "t": 2.981673240661621}
|
| 205 |
+
{"step": 205, "loss": 1.110738754272461, "grad_norm": 0.5835689306259155, "lr": 2.8922540968615286e-05, "epoch": 1, "t": 3.119799852371216}
|
| 206 |
+
{"step": 206, "loss": 1.15838623046875, "grad_norm": 0.5934703946113586, "lr": 2.8442518621591086e-05, "epoch": 1, "t": 2.9539198875427246}
|
| 207 |
+
{"step": 207, "loss": 1.1605558395385742, "grad_norm": 0.6126264929771423, "lr": 2.7964924618885778e-05, "epoch": 1, "t": 3.270961284637451}
|
| 208 |
+
{"step": 208, "loss": 1.1961784362792969, "grad_norm": 0.6048758029937744, "lr": 2.748981275911633e-05, "epoch": 1, "t": 3.294602394104004}
|
| 209 |
+
{"step": 209, "loss": 1.1333205699920654, "grad_norm": 0.6265262961387634, "lr": 2.701723656129851e-05, "epoch": 1, "t": 2.9729597568511963}
|
| 210 |
+
{"step": 210, "loss": 1.1624705791473389, "grad_norm": 0.5679384469985962, "lr": 2.6547249258818164e-05, "epoch": 1, "t": 3.3164453506469727}
|
| 211 |
+
{"step": 211, "loss": 1.0498530864715576, "grad_norm": 0.6415156722068787, "lr": 2.607990379343489e-05, "epoch": 1, "t": 2.941053628921509}
|
| 212 |
+
{"step": 212, "loss": 1.018575668334961, "grad_norm": 0.6429729461669922, "lr": 2.5615252809318284e-05, "epoch": 1, "t": 2.953709602355957}
|
| 213 |
+
{"step": 213, "loss": 1.295606255531311, "grad_norm": 0.5761866569519043, "lr": 2.5153348647117857e-05, "epoch": 1, "t": 3.5970351696014404}
|
| 214 |
+
{"step": 214, "loss": 1.1575409173965454, "grad_norm": 0.5509631037712097, "lr": 2.469424333806718e-05, "epoch": 1, "t": 3.4427781105041504}
|
| 215 |
+
{"step": 215, "loss": 1.046026349067688, "grad_norm": 0.626361608505249, "lr": 2.4237988598122752e-05, "epoch": 1, "t": 3.0473265647888184}
|
| 216 |
+
{"step": 216, "loss": 1.2087295055389404, "grad_norm": 0.6969119310379028, "lr": 2.3784635822138424e-05, "epoch": 1, "t": 3.1313812732696533}
|
| 217 |
+
{"step": 217, "loss": 1.1948926448822021, "grad_norm": 0.5655863881111145, "lr": 2.333423607807613e-05, "epoch": 1, "t": 3.246502161026001}
|
| 218 |
+
{"step": 218, "loss": 1.185545802116394, "grad_norm": 0.6220167279243469, "lr": 2.288684010125325e-05, "epoch": 1, "t": 3.1777689456939697}
|
| 219 |
+
{"step": 219, "loss": 1.1585865020751953, "grad_norm": 0.6115987300872803, "lr": 2.2442498288627556e-05, "epoch": 1, "t": 3.0028374195098877}
|
| 220 |
+
{"step": 220, "loss": 1.045722246170044, "grad_norm": 0.652891993522644, "lr": 2.2001260693120233e-05, "epoch": 1, "t": 3.1166067123413086}
|
| 221 |
+
{"step": 221, "loss": 1.0521082878112793, "grad_norm": 0.6072604060173035, "lr": 2.156317701797766e-05, "epoch": 1, "t": 2.929232120513916}
|
| 222 |
+
{"step": 222, "loss": 1.1327838897705078, "grad_norm": 0.5822096467018127, "lr": 2.1128296611172593e-05, "epoch": 1, "t": 3.2047457695007324}
|
| 223 |
+
{"step": 223, "loss": 1.1181288957595825, "grad_norm": 0.665998101234436, "lr": 2.0696668459845355e-05, "epoch": 1, "t": 2.911134719848633}
|
| 224 |
+
{"step": 224, "loss": 1.2542016506195068, "grad_norm": 0.6517634987831116, "lr": 2.026834118478567e-05, "epoch": 1, "t": 3.3153140544891357}
|
| 225 |
+
{"step": 225, "loss": 1.1514021158218384, "grad_norm": 0.6453463435173035, "lr": 1.98433630349558e-05, "epoch": 1, "t": 3.173335075378418}
|
| 226 |
+
{"step": 226, "loss": 1.0198673009872437, "grad_norm": 0.6537891626358032, "lr": 1.9421781882055444e-05, "epoch": 1, "t": 2.895737648010254}
|
| 227 |
+
{"step": 227, "loss": 1.1016989946365356, "grad_norm": 0.6124733090400696, "lr": 1.9003645215129355e-05, "epoch": 1, "t": 3.137103796005249}
|
| 228 |
+
{"step": 228, "loss": 1.2772940397262573, "grad_norm": 0.6474252343177795, "lr": 1.858900013521788e-05, "epoch": 1, "t": 3.2814319133758545}
|
| 229 |
+
{"step": 229, "loss": 1.0964081287384033, "grad_norm": 0.5437510013580322, "lr": 1.817789335005121e-05, "epoch": 1, "t": 3.5712926387786865}
|
| 230 |
+
{"step": 230, "loss": 1.2341210842132568, "grad_norm": 0.5824730396270752, "lr": 1.777037116878804e-05, "epoch": 1, "t": 3.067516803741455}
|
| 231 |
+
{"step": 231, "loss": 1.1912215948104858, "grad_norm": 0.6221261024475098, "lr": 1.7366479496799077e-05, "epoch": 1, "t": 3.298513412475586}
|
| 232 |
+
{"step": 232, "loss": 1.171180009841919, "grad_norm": 0.6992931365966797, "lr": 1.6966263830495936e-05, "epoch": 1, "t": 2.959256172180176}
|
| 233 |
+
{"step": 233, "loss": 1.107611894607544, "grad_norm": 0.5788571834564209, "lr": 1.656976925220633e-05, "epoch": 1, "t": 3.1884186267852783}
|
| 234 |
+
{"step": 234, "loss": 1.1315503120422363, "grad_norm": 0.5943876504898071, "lr": 1.6177040425095662e-05, "epoch": 1, "t": 3.4186887741088867}
|
| 235 |
+
{"step": 235, "loss": 1.2423261404037476, "grad_norm": 0.608063817024231, "lr": 1.5788121588135975e-05, "epoch": 1, "t": 2.972113847732544}
|
| 236 |
+
{"step": 236, "loss": 1.1326618194580078, "grad_norm": 0.6260091066360474, "lr": 1.5403056551122697e-05, "epoch": 1, "t": 3.172067403793335}
|
| 237 |
+
{"step": 237, "loss": 1.1549441814422607, "grad_norm": 0.5190547108650208, "lr": 1.5021888689739549e-05, "epoch": 1, "t": 3.217885971069336}
|
| 238 |
+
{"step": 238, "loss": 1.2121508121490479, "grad_norm": 0.6164082288742065, "lr": 1.4644660940672627e-05, "epoch": 1, "t": 3.116456985473633}
|
| 239 |
+
{"step": 239, "loss": 1.186787486076355, "grad_norm": 0.7234702110290527, "lr": 1.427141579677374e-05, "epoch": 1, "t": 3.143021583557129}
|
| 240 |
+
{"step": 240, "loss": 1.0911130905151367, "grad_norm": 0.5621363520622253, "lr": 1.3902195302273779e-05, "epoch": 1, "t": 3.230220317840576}
|
| 241 |
+
{"step": 241, "loss": 1.1037520170211792, "grad_norm": 0.5908945798873901, "lr": 1.3537041048046695e-05, "epoch": 1, "t": 3.3280298709869385}
|
| 242 |
+
{"step": 242, "loss": 1.1579039096832275, "grad_norm": 0.6530472040176392, "lr": 1.3175994166924394e-05, "epoch": 1, "t": 3.017481565475464}
|
| 243 |
+
{"step": 243, "loss": 1.1008596420288086, "grad_norm": 0.5923284888267517, "lr": 1.2819095329063469e-05, "epoch": 1, "t": 2.8944971561431885}
|
| 244 |
+
{"step": 244, "loss": 1.0920989513397217, "grad_norm": 0.6365619897842407, "lr": 1.246638473736378e-05, "epoch": 1, "t": 2.98527193069458}
|
| 245 |
+
{"step": 245, "loss": 1.150795817375183, "grad_norm": 0.5903105139732361, "lr": 1.2117902122939861e-05, "epoch": 1, "t": 3.351747751235962}
|
| 246 |
+
{"step": 246, "loss": 1.1814042329788208, "grad_norm": 0.6527220010757446, "lr": 1.1773686740645384e-05, "epoch": 1, "t": 3.2993481159210205}
|
| 247 |
+
{"step": 247, "loss": 1.041551947593689, "grad_norm": 0.6108824014663696, "lr": 1.1433777364651271e-05, "epoch": 1, "t": 2.9131946563720703}
|
| 248 |
+
{"step": 248, "loss": 0.9981467127799988, "grad_norm": 0.72492915391922, "lr": 1.1098212284078036e-05, "epoch": 1, "t": 3.004458427429199}
|
| 249 |
+
{"step": 249, "loss": 1.0924677848815918, "grad_norm": 0.642476499080658, "lr": 1.076702929868264e-05, "epoch": 1, "t": 3.0539562702178955}
|
| 250 |
+
{"step": 250, "loss": 1.0307005643844604, "grad_norm": 0.5619496703147888, "lr": 1.0440265714600572e-05, "epoch": 1, "t": 3.3310062885284424}
|
| 251 |
+
{"step": 251, "loss": 1.143161416053772, "grad_norm": 0.5786762237548828, "lr": 1.0117958340143507e-05, "epoch": 1, "t": 3.5837273597717285}
|
| 252 |
+
{"step": 252, "loss": 1.2036066055297852, "grad_norm": 0.5805399417877197, "lr": 9.800143481652979e-06, "epoch": 1, "t": 3.4024734497070312}
|
| 253 |
+
{"step": 253, "loss": 1.2250564098358154, "grad_norm": 0.6205870509147644, "lr": 9.48685693941067e-06, "epoch": 1, "t": 3.296753168106079}
|
| 254 |
+
{"step": 254, "loss": 1.1756951808929443, "grad_norm": 0.5823423266410828, "lr": 9.17813400360572e-06, "epoch": 1, "t": 3.418215036392212}
|
| 255 |
+
{"step": 255, "loss": 1.1648709774017334, "grad_norm": 0.577837347984314, "lr": 8.874009450359427e-06, "epoch": 1, "t": 3.1661248207092285}
|
| 256 |
+
{"step": 256, "loss": 1.0923452377319336, "grad_norm": 0.5667760968208313, "lr": 8.574517537807897e-06, "epoch": 1, "t": 3.2927653789520264}
|
| 257 |
+
{"step": 257, "loss": 1.133958101272583, "grad_norm": 0.6322847008705139, "lr": 8.279692002243027e-06, "epoch": 1, "t": 3.098123550415039}
|
| 258 |
+
{"step": 258, "loss": 1.1190788745880127, "grad_norm": 0.5409640669822693, "lr": 7.989566054312287e-06, "epoch": 1, "t": 3.3374855518341064}
|
| 259 |
+
{"step": 259, "loss": 1.0914477109909058, "grad_norm": 0.582192063331604, "lr": 7.704172375277691e-06, "epoch": 1, "t": 3.106818437576294}
|
| 260 |
+
{"step": 260, "loss": 1.1423075199127197, "grad_norm": 0.6024892330169678, "lr": 7.423543113334436e-06, "epoch": 1, "t": 3.0023627281188965}
|
| 261 |
+
{"step": 261, "loss": 1.0804548263549805, "grad_norm": 0.630786120891571, "lr": 7.14770987998954e-06, "epoch": 1, "t": 3.145784616470337}
|
| 262 |
+
{"step": 262, "loss": 1.0986104011535645, "grad_norm": 0.5946423411369324, "lr": 6.876703746500984e-06, "epoch": 1, "t": 3.0009729862213135}
|
| 263 |
+
{"step": 263, "loss": 1.1484415531158447, "grad_norm": 0.6121066212654114, "lr": 6.610555240377652e-06, "epoch": 1, "t": 3.253079414367676}
|
| 264 |
+
{"step": 264, "loss": 1.1844849586486816, "grad_norm": 0.6580023169517517, "lr": 6.349294341940593e-06, "epoch": 1, "t": 2.939393997192383}
|
| 265 |
+
{"step": 265, "loss": 1.2188023328781128, "grad_norm": 0.6581462621688843, "lr": 6.092950480945897e-06, "epoch": 1, "t": 3.072070360183716}
|
| 266 |
+
{"step": 266, "loss": 1.162814736366272, "grad_norm": 0.5939738750457764, "lr": 5.841552533269534e-06, "epoch": 1, "t": 3.5493392944335938}
|
| 267 |
+
{"step": 267, "loss": 1.2604751586914062, "grad_norm": 0.6394745111465454, "lr": 5.595128817654638e-06, "epoch": 1, "t": 3.4884896278381348}
|
| 268 |
+
{"step": 268, "loss": 1.146340250968933, "grad_norm": 0.52507084608078, "lr": 5.353707092521582e-06, "epoch": 1, "t": 3.546063184738159}
|
| 269 |
+
{"step": 269, "loss": 1.1483925580978394, "grad_norm": 0.6097657084465027, "lr": 5.117314552841052e-06, "epoch": 1, "t": 3.0640063285827637}
|
| 270 |
+
{"step": 270, "loss": 1.1280434131622314, "grad_norm": 0.6271836161613464, "lr": 4.885977827070748e-06, "epoch": 1, "t": 3.3846919536590576}
|
| 271 |
+
{"step": 271, "loss": 1.1444844007492065, "grad_norm": 0.6049522161483765, "lr": 4.659722974155767e-06, "epoch": 1, "t": 2.9123172760009766}
|
| 272 |
+
{"step": 272, "loss": 1.21195387840271, "grad_norm": 0.6936115026473999, "lr": 4.43857548059321e-06, "epoch": 1, "t": 3.290844440460205}
|
| 273 |
+
{"step": 273, "loss": 1.083553433418274, "grad_norm": 0.5946942567825317, "lr": 4.2225602575612755e-06, "epoch": 1, "t": 3.0923309326171875}
|
| 274 |
+
{"step": 274, "loss": 1.1764330863952637, "grad_norm": 0.6581479907035828, "lr": 4.011701638113063e-06, "epoch": 1, "t": 2.9411404132843018}
|
| 275 |
+
{"step": 275, "loss": 1.020877718925476, "grad_norm": 0.6329962611198425, "lr": 3.8060233744356633e-06, "epoch": 1, "t": 3.289860486984253}
|
| 276 |
+
{"step": 276, "loss": 1.246074914932251, "grad_norm": 0.6978022456169128, "lr": 3.605548635174533e-06, "epoch": 1, "t": 3.033954381942749}
|
| 277 |
+
{"step": 277, "loss": 1.1520459651947021, "grad_norm": 0.5991973280906677, "lr": 3.410300002823691e-06, "epoch": 1, "t": 3.0740439891815186}
|
| 278 |
+
{"step": 278, "loss": 1.107762336730957, "grad_norm": 0.6086584329605103, "lr": 3.220299471181898e-06, "epoch": 1, "t": 3.1639225482940674}
|
| 279 |
+
{"step": 279, "loss": 1.185377836227417, "grad_norm": 0.5857842564582825, "lr": 3.035568442875136e-06, "epoch": 1, "t": 3.148939847946167}
|
| 280 |
+
{"step": 280, "loss": 1.1530919075012207, "grad_norm": 0.620843768119812, "lr": 2.85612772694579e-06, "epoch": 1, "t": 3.309561014175415}
|
| 281 |
+
{"step": 281, "loss": 1.1242648363113403, "grad_norm": 0.596558153629303, "lr": 2.6819975365085237e-06, "epoch": 1, "t": 2.996490240097046}
|
| 282 |
+
{"step": 282, "loss": 1.1723312139511108, "grad_norm": 0.5756584405899048, "lr": 2.5131974864734066e-06, "epoch": 1, "t": 3.3760790824890137}
|
| 283 |
+
{"step": 283, "loss": 1.1036540269851685, "grad_norm": 0.5501166582107544, "lr": 2.349746591336405e-06, "epoch": 1, "t": 3.4217169284820557}
|
| 284 |
+
{"step": 284, "loss": 1.2204954624176025, "grad_norm": 0.5722854733467102, "lr": 2.191663263037458e-06, "epoch": 1, "t": 3.1736488342285156}
|
| 285 |
+
{"step": 285, "loss": 1.0018733739852905, "grad_norm": 0.5935145616531372, "lr": 2.0389653088865036e-06, "epoch": 1, "t": 3.4486806392669678}
|
| 286 |
+
{"step": 286, "loss": 1.10654616355896, "grad_norm": 0.6606000661849976, "lr": 1.8916699295575324e-06, "epoch": 1, "t": 3.0607917308807373}
|
| 287 |
+
{"step": 287, "loss": 1.1920475959777832, "grad_norm": 0.6654186248779297, "lr": 1.7497937171510547e-06, "epoch": 1, "t": 3.1877105236053467}
|
| 288 |
+
{"step": 288, "loss": 1.1609740257263184, "grad_norm": 0.5407600998878479, "lr": 1.6133526533250565e-06, "epoch": 1, "t": 3.0744001865386963}
|
| 289 |
+
{"step": 289, "loss": 1.1973578929901123, "grad_norm": 0.5786117315292358, "lr": 1.4823621074947503e-06, "epoch": 1, "t": 3.437222719192505}
|
| 290 |
+
{"step": 290, "loss": 1.2084985971450806, "grad_norm": 0.6255794763565063, "lr": 1.3568368351012717e-06, "epoch": 1, "t": 3.1683363914489746}
|
| 291 |
+
{"step": 291, "loss": 1.0135982036590576, "grad_norm": 0.5914056301116943, "lr": 1.236790975949592e-06, "epoch": 1, "t": 3.0157432556152344}
|
| 292 |
+
{"step": 292, "loss": 1.1606616973876953, "grad_norm": 0.6107684373855591, "lr": 1.1222380526156928e-06, "epoch": 1, "t": 3.106762170791626}
|
| 293 |
+
{"step": 293, "loss": 1.0304497480392456, "grad_norm": 0.5566821098327637, "lr": 1.0131909689233442e-06, "epoch": 1, "t": 3.210559129714966}
|
| 294 |
+
{"step": 294, "loss": 1.201225757598877, "grad_norm": 0.5820565223693848, "lr": 9.096620084905472e-07, "epoch": 1, "t": 3.1978530883789062}
|
| 295 |
+
{"step": 295, "loss": 1.1365504264831543, "grad_norm": 0.560297966003418, "lr": 8.11662833345822e-07, "epoch": 1, "t": 3.4149467945098877}
|
| 296 |
+
{"step": 296, "loss": 1.2560802698135376, "grad_norm": 0.732952356338501, "lr": 7.192044826145771e-07, "epoch": 1, "t": 3.2335638999938965}
|
| 297 |
+
{"step": 297, "loss": 1.0901637077331543, "grad_norm": 0.6319404244422913, "lr": 6.322973712755697e-07, "epoch": 1, "t": 3.2433555126190186}
|
| 298 |
+
{"step": 298, "loss": 1.147761583328247, "grad_norm": 0.5701617002487183, "lr": 5.509512889877333e-07, "epoch": 1, "t": 3.0195932388305664}
|
| 299 |
+
{"step": 299, "loss": 1.1304981708526611, "grad_norm": 0.6197502613067627, "lr": 4.7517539898741524e-07, "epoch": 1, "t": 2.9945428371429443}
|
| 300 |
+
{"step": 300, "loss": 1.1311135292053223, "grad_norm": 0.5890395641326904, "lr": 4.049782370561583e-07, "epoch": 1, "t": 3.0199780464172363}
|
| 301 |
+
{"step": 301, "loss": 1.185089349746704, "grad_norm": 0.6514334082603455, "lr": 3.4036771055923066e-07, "epoch": 1, "t": 3.4238433837890625}
|
| 302 |
+
{"step": 302, "loss": 1.2313034534454346, "grad_norm": 0.6727525591850281, "lr": 2.813510975548772e-07, "epoch": 1, "t": 2.967966318130493}
|
| 303 |
+
{"step": 303, "loss": 1.0683801174163818, "grad_norm": 0.6761717796325684, "lr": 2.2793504597447002e-07, "epoch": 1, "t": 3.156531572341919}
|
| 304 |
+
{"step": 304, "loss": 1.1297897100448608, "grad_norm": 0.6605789065361023, "lr": 1.8012557287367392e-07, "epoch": 1, "t": 3.8942816257476807}
|
| 305 |
+
{"step": 305, "loss": 1.1612855195999146, "grad_norm": 0.5988486409187317, "lr": 1.379280637546443e-07, "epoch": 1, "t": 3.2300262451171875}
|
| 306 |
+
{"step": 306, "loss": 1.149768352508545, "grad_norm": 0.575775682926178, "lr": 1.0134727195937333e-07, "epoch": 1, "t": 3.4097437858581543}
|
| 307 |
+
{"step": 307, "loss": 1.2025296688079834, "grad_norm": 0.6404022574424744, "lr": 7.038731813426291e-08, "epoch": 1, "t": 3.135118246078491}
|
| 308 |
+
{"step": 308, "loss": 1.2290019989013672, "grad_norm": 0.6150823831558228, "lr": 4.5051689765929214e-08, "epoch": 1, "t": 3.4961788654327393}
|
| 309 |
+
{"step": 309, "loss": 1.0409400463104248, "grad_norm": 0.5782963633537292, "lr": 2.534324078837802e-08, "epoch": 1, "t": 3.5864033699035645}
|
| 310 |
+
{"step": 310, "loss": 1.1258490085601807, "grad_norm": 0.5588079690933228, "lr": 1.1264191261528557e-08, "epoch": 1, "t": 3.306751012802124}
|
| 311 |
+
{"step": 311, "loss": 1.0926461219787598, "grad_norm": 0.662872850894928, "lr": 2.8161271211024633e-09, "epoch": 1, "t": 3.037998914718628}
|
| 312 |
+
{"step": 312, "loss": 1.1257848739624023, "grad_norm": 0.6854902505874634, "lr": 0.0, "epoch": 1, "t": 3.279524803161621}
|
12b/u_bottom-s0/README.md
ADDED
|
@@ -0,0 +1,206 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: google/gemma-3-12b-it
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- base_model:adapter:google/gemma-3-12b-it
|
| 6 |
+
- lora
|
| 7 |
+
- transformers
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# Model Card for Model ID
|
| 11 |
+
|
| 12 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
## Model Details
|
| 17 |
+
|
| 18 |
+
### Model Description
|
| 19 |
+
|
| 20 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
- **Developed by:** [More Information Needed]
|
| 25 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 26 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 27 |
+
- **Model type:** [More Information Needed]
|
| 28 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 29 |
+
- **License:** [More Information Needed]
|
| 30 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 31 |
+
|
| 32 |
+
### Model Sources [optional]
|
| 33 |
+
|
| 34 |
+
<!-- Provide the basic links for the model. -->
|
| 35 |
+
|
| 36 |
+
- **Repository:** [More Information Needed]
|
| 37 |
+
- **Paper [optional]:** [More Information Needed]
|
| 38 |
+
- **Demo [optional]:** [More Information Needed]
|
| 39 |
+
|
| 40 |
+
## Uses
|
| 41 |
+
|
| 42 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 43 |
+
|
| 44 |
+
### Direct Use
|
| 45 |
+
|
| 46 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 47 |
+
|
| 48 |
+
[More Information Needed]
|
| 49 |
+
|
| 50 |
+
### Downstream Use [optional]
|
| 51 |
+
|
| 52 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 53 |
+
|
| 54 |
+
[More Information Needed]
|
| 55 |
+
|
| 56 |
+
### Out-of-Scope Use
|
| 57 |
+
|
| 58 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 59 |
+
|
| 60 |
+
[More Information Needed]
|
| 61 |
+
|
| 62 |
+
## Bias, Risks, and Limitations
|
| 63 |
+
|
| 64 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 65 |
+
|
| 66 |
+
[More Information Needed]
|
| 67 |
+
|
| 68 |
+
### Recommendations
|
| 69 |
+
|
| 70 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 71 |
+
|
| 72 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 73 |
+
|
| 74 |
+
## How to Get Started with the Model
|
| 75 |
+
|
| 76 |
+
Use the code below to get started with the model.
|
| 77 |
+
|
| 78 |
+
[More Information Needed]
|
| 79 |
+
|
| 80 |
+
## Training Details
|
| 81 |
+
|
| 82 |
+
### Training Data
|
| 83 |
+
|
| 84 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 85 |
+
|
| 86 |
+
[More Information Needed]
|
| 87 |
+
|
| 88 |
+
### Training Procedure
|
| 89 |
+
|
| 90 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 91 |
+
|
| 92 |
+
#### Preprocessing [optional]
|
| 93 |
+
|
| 94 |
+
[More Information Needed]
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
#### Training Hyperparameters
|
| 98 |
+
|
| 99 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 100 |
+
|
| 101 |
+
#### Speeds, Sizes, Times [optional]
|
| 102 |
+
|
| 103 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 104 |
+
|
| 105 |
+
[More Information Needed]
|
| 106 |
+
|
| 107 |
+
## Evaluation
|
| 108 |
+
|
| 109 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 110 |
+
|
| 111 |
+
### Testing Data, Factors & Metrics
|
| 112 |
+
|
| 113 |
+
#### Testing Data
|
| 114 |
+
|
| 115 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 116 |
+
|
| 117 |
+
[More Information Needed]
|
| 118 |
+
|
| 119 |
+
#### Factors
|
| 120 |
+
|
| 121 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 122 |
+
|
| 123 |
+
[More Information Needed]
|
| 124 |
+
|
| 125 |
+
#### Metrics
|
| 126 |
+
|
| 127 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 128 |
+
|
| 129 |
+
[More Information Needed]
|
| 130 |
+
|
| 131 |
+
### Results
|
| 132 |
+
|
| 133 |
+
[More Information Needed]
|
| 134 |
+
|
| 135 |
+
#### Summary
|
| 136 |
+
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
## Model Examination [optional]
|
| 140 |
+
|
| 141 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 142 |
+
|
| 143 |
+
[More Information Needed]
|
| 144 |
+
|
| 145 |
+
## Environmental Impact
|
| 146 |
+
|
| 147 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 148 |
+
|
| 149 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 150 |
+
|
| 151 |
+
- **Hardware Type:** [More Information Needed]
|
| 152 |
+
- **Hours used:** [More Information Needed]
|
| 153 |
+
- **Cloud Provider:** [More Information Needed]
|
| 154 |
+
- **Compute Region:** [More Information Needed]
|
| 155 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 156 |
+
|
| 157 |
+
## Technical Specifications [optional]
|
| 158 |
+
|
| 159 |
+
### Model Architecture and Objective
|
| 160 |
+
|
| 161 |
+
[More Information Needed]
|
| 162 |
+
|
| 163 |
+
### Compute Infrastructure
|
| 164 |
+
|
| 165 |
+
[More Information Needed]
|
| 166 |
+
|
| 167 |
+
#### Hardware
|
| 168 |
+
|
| 169 |
+
[More Information Needed]
|
| 170 |
+
|
| 171 |
+
#### Software
|
| 172 |
+
|
| 173 |
+
[More Information Needed]
|
| 174 |
+
|
| 175 |
+
## Citation [optional]
|
| 176 |
+
|
| 177 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 178 |
+
|
| 179 |
+
**BibTeX:**
|
| 180 |
+
|
| 181 |
+
[More Information Needed]
|
| 182 |
+
|
| 183 |
+
**APA:**
|
| 184 |
+
|
| 185 |
+
[More Information Needed]
|
| 186 |
+
|
| 187 |
+
## Glossary [optional]
|
| 188 |
+
|
| 189 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 190 |
+
|
| 191 |
+
[More Information Needed]
|
| 192 |
+
|
| 193 |
+
## More Information [optional]
|
| 194 |
+
|
| 195 |
+
[More Information Needed]
|
| 196 |
+
|
| 197 |
+
## Model Card Authors [optional]
|
| 198 |
+
|
| 199 |
+
[More Information Needed]
|
| 200 |
+
|
| 201 |
+
## Model Card Contact
|
| 202 |
+
|
| 203 |
+
[More Information Needed]
|
| 204 |
+
### Framework versions
|
| 205 |
+
|
| 206 |
+
- PEFT 0.17.1
|
12b/u_bottom-s0/adapter_config.json
ADDED
|
@@ -0,0 +1,186 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": {
|
| 4 |
+
"base_model_class": "Gemma3ForConditionalGeneration",
|
| 5 |
+
"parent_library": "transformers.models.gemma3.modeling_gemma3"
|
| 6 |
+
},
|
| 7 |
+
"base_model_name_or_path": "google/gemma-3-12b-it",
|
| 8 |
+
"bias": "none",
|
| 9 |
+
"corda_config": null,
|
| 10 |
+
"eva_config": null,
|
| 11 |
+
"exclude_modules": null,
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": true,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layer_replication": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"layers_to_transform": null,
|
| 18 |
+
"loftq_config": {},
|
| 19 |
+
"lora_alpha": 128,
|
| 20 |
+
"lora_bias": false,
|
| 21 |
+
"lora_dropout": 0.05,
|
| 22 |
+
"megatron_config": null,
|
| 23 |
+
"megatron_core": "megatron.core",
|
| 24 |
+
"modules_to_save": null,
|
| 25 |
+
"peft_type": "LORA",
|
| 26 |
+
"qalora_group_size": 16,
|
| 27 |
+
"r": 64,
|
| 28 |
+
"rank_pattern": {},
|
| 29 |
+
"revision": null,
|
| 30 |
+
"target_modules": [
|
| 31 |
+
"28.self_attn.k_proj",
|
| 32 |
+
"language_model.layers.13.self_attn.k_proj",
|
| 33 |
+
"language_model.layers.18.self_attn.k_proj",
|
| 34 |
+
"language_model.layers.4.self_attn.k_proj",
|
| 35 |
+
"language_model.layers.10.self_attn.k_proj",
|
| 36 |
+
"language_model.layers.6.self_attn.v_proj",
|
| 37 |
+
"45.self_attn.q_proj",
|
| 38 |
+
"27.self_attn.v_proj",
|
| 39 |
+
"language_model.layers.11.self_attn.v_proj",
|
| 40 |
+
"language_model.layers.9.self_attn.k_proj",
|
| 41 |
+
"language_model.layers.2.self_attn.q_proj",
|
| 42 |
+
"34.self_attn.q_proj",
|
| 43 |
+
"28.self_attn.v_proj",
|
| 44 |
+
"44.self_attn.k_proj",
|
| 45 |
+
"language_model.layers.2.self_attn.k_proj",
|
| 46 |
+
"language_model.layers.17.self_attn.v_proj",
|
| 47 |
+
"34.self_attn.k_proj",
|
| 48 |
+
"42.self_attn.v_proj",
|
| 49 |
+
"language_model.layers.24.self_attn.k_proj",
|
| 50 |
+
"language_model.layers.26.self_attn.v_proj",
|
| 51 |
+
"30.self_attn.q_proj",
|
| 52 |
+
"27.self_attn.k_proj",
|
| 53 |
+
"41.self_attn.q_proj",
|
| 54 |
+
"language_model.layers.24.self_attn.q_proj",
|
| 55 |
+
"language_model.layers.2.self_attn.v_proj",
|
| 56 |
+
"language_model.layers.11.self_attn.k_proj",
|
| 57 |
+
"45.self_attn.k_proj",
|
| 58 |
+
"39.self_attn.q_proj",
|
| 59 |
+
"language_model.layers.14.self_attn.q_proj",
|
| 60 |
+
"language_model.layers.10.self_attn.v_proj",
|
| 61 |
+
"35.self_attn.v_proj",
|
| 62 |
+
"language_model.layers.23.self_attn.k_proj",
|
| 63 |
+
"language_model.layers.8.self_attn.q_proj",
|
| 64 |
+
"language_model.layers.9.self_attn.q_proj",
|
| 65 |
+
"language_model.layers.9.self_attn.v_proj",
|
| 66 |
+
"language_model.layers.13.self_attn.v_proj",
|
| 67 |
+
"language_model.layers.5.self_attn.v_proj",
|
| 68 |
+
"language_model.layers.3.self_attn.q_proj",
|
| 69 |
+
"31.self_attn.v_proj",
|
| 70 |
+
"language_model.layers.22.self_attn.k_proj",
|
| 71 |
+
"language_model.layers.4.self_attn.v_proj",
|
| 72 |
+
"language_model.layers.12.self_attn.v_proj",
|
| 73 |
+
"language_model.layers.16.self_attn.k_proj",
|
| 74 |
+
"language_model.layers.20.self_attn.k_proj",
|
| 75 |
+
"language_model.layers.25.self_attn.q_proj",
|
| 76 |
+
"language_model.layers.18.self_attn.q_proj",
|
| 77 |
+
"language_model.layers.4.self_attn.q_proj",
|
| 78 |
+
"language_model.layers.15.self_attn.q_proj",
|
| 79 |
+
"41.self_attn.v_proj",
|
| 80 |
+
"language_model.layers.21.self_attn.k_proj",
|
| 81 |
+
"language_model.layers.21.self_attn.q_proj",
|
| 82 |
+
"language_model.layers.23.self_attn.q_proj",
|
| 83 |
+
"43.self_attn.k_proj",
|
| 84 |
+
"up_proj",
|
| 85 |
+
"47.self_attn.q_proj",
|
| 86 |
+
"language_model.layers.5.self_attn.k_proj",
|
| 87 |
+
"30.self_attn.v_proj",
|
| 88 |
+
"language_model.layers.7.self_attn.v_proj",
|
| 89 |
+
"29.self_attn.k_proj",
|
| 90 |
+
"language_model.layers.0.self_attn.v_proj",
|
| 91 |
+
"o_proj",
|
| 92 |
+
"33.self_attn.q_proj",
|
| 93 |
+
"language_model.layers.15.self_attn.k_proj",
|
| 94 |
+
"language_model.layers.6.self_attn.q_proj",
|
| 95 |
+
"language_model.layers.24.self_attn.v_proj",
|
| 96 |
+
"language_model.layers.20.self_attn.v_proj",
|
| 97 |
+
"language_model.layers.25.self_attn.v_proj",
|
| 98 |
+
"language_model.layers.25.self_attn.k_proj",
|
| 99 |
+
"46.self_attn.q_proj",
|
| 100 |
+
"31.self_attn.q_proj",
|
| 101 |
+
"29.self_attn.q_proj",
|
| 102 |
+
"language_model.layers.12.self_attn.k_proj",
|
| 103 |
+
"39.self_attn.v_proj",
|
| 104 |
+
"language_model.layers.17.self_attn.q_proj",
|
| 105 |
+
"language_model.layers.8.self_attn.k_proj",
|
| 106 |
+
"43.self_attn.q_proj",
|
| 107 |
+
"32.self_attn.k_proj",
|
| 108 |
+
"39.self_attn.k_proj",
|
| 109 |
+
"language_model.layers.16.self_attn.q_proj",
|
| 110 |
+
"language_model.layers.21.self_attn.v_proj",
|
| 111 |
+
"language_model.layers.18.self_attn.v_proj",
|
| 112 |
+
"38.self_attn.v_proj",
|
| 113 |
+
"33.self_attn.k_proj",
|
| 114 |
+
"language_model.layers.13.self_attn.q_proj",
|
| 115 |
+
"language_model.layers.23.self_attn.v_proj",
|
| 116 |
+
"language_model.layers.0.self_attn.q_proj",
|
| 117 |
+
"language_model.layers.3.self_attn.k_proj",
|
| 118 |
+
"language_model.layers.6.self_attn.k_proj",
|
| 119 |
+
"27.self_attn.q_proj",
|
| 120 |
+
"language_model.layers.26.self_attn.q_proj",
|
| 121 |
+
"47.self_attn.v_proj",
|
| 122 |
+
"40.self_attn.k_proj",
|
| 123 |
+
"32.self_attn.q_proj",
|
| 124 |
+
"language_model.layers.1.self_attn.v_proj",
|
| 125 |
+
"language_model.layers.7.self_attn.k_proj",
|
| 126 |
+
"language_model.layers.0.self_attn.k_proj",
|
| 127 |
+
"gate_proj",
|
| 128 |
+
"28.self_attn.q_proj",
|
| 129 |
+
"42.self_attn.k_proj",
|
| 130 |
+
"41.self_attn.k_proj",
|
| 131 |
+
"44.self_attn.q_proj",
|
| 132 |
+
"language_model.layers.19.self_attn.k_proj",
|
| 133 |
+
"language_model.layers.14.self_attn.v_proj",
|
| 134 |
+
"language_model.layers.22.self_attn.v_proj",
|
| 135 |
+
"language_model.layers.20.self_attn.q_proj",
|
| 136 |
+
"language_model.layers.17.self_attn.k_proj",
|
| 137 |
+
"47.self_attn.k_proj",
|
| 138 |
+
"down_proj",
|
| 139 |
+
"37.self_attn.q_proj",
|
| 140 |
+
"language_model.layers.15.self_attn.v_proj",
|
| 141 |
+
"language_model.layers.19.self_attn.v_proj",
|
| 142 |
+
"34.self_attn.v_proj",
|
| 143 |
+
"35.self_attn.q_proj",
|
| 144 |
+
"40.self_attn.v_proj",
|
| 145 |
+
"36.self_attn.v_proj",
|
| 146 |
+
"35.self_attn.k_proj",
|
| 147 |
+
"30.self_attn.k_proj",
|
| 148 |
+
"language_model.layers.7.self_attn.q_proj",
|
| 149 |
+
"language_model.layers.1.self_attn.q_proj",
|
| 150 |
+
"37.self_attn.v_proj",
|
| 151 |
+
"language_model.layers.10.self_attn.q_proj",
|
| 152 |
+
"language_model.layers.8.self_attn.v_proj",
|
| 153 |
+
"language_model.layers.3.self_attn.v_proj",
|
| 154 |
+
"36.self_attn.q_proj",
|
| 155 |
+
"language_model.layers.11.self_attn.q_proj",
|
| 156 |
+
"language_model.layers.12.self_attn.q_proj",
|
| 157 |
+
"40.self_attn.q_proj",
|
| 158 |
+
"42.self_attn.q_proj",
|
| 159 |
+
"language_model.layers.19.self_attn.q_proj",
|
| 160 |
+
"32.self_attn.v_proj",
|
| 161 |
+
"37.self_attn.k_proj",
|
| 162 |
+
"31.self_attn.k_proj",
|
| 163 |
+
"language_model.layers.1.self_attn.k_proj",
|
| 164 |
+
"language_model.layers.22.self_attn.q_proj",
|
| 165 |
+
"38.self_attn.q_proj",
|
| 166 |
+
"45.self_attn.v_proj",
|
| 167 |
+
"46.self_attn.k_proj",
|
| 168 |
+
"33.self_attn.v_proj",
|
| 169 |
+
"language_model.layers.16.self_attn.v_proj",
|
| 170 |
+
"language_model.layers.14.self_attn.k_proj",
|
| 171 |
+
"36.self_attn.k_proj",
|
| 172 |
+
"language_model.layers.5.self_attn.q_proj",
|
| 173 |
+
"language_model.layers.26.self_attn.k_proj",
|
| 174 |
+
"38.self_attn.k_proj",
|
| 175 |
+
"46.self_attn.v_proj",
|
| 176 |
+
"43.self_attn.v_proj",
|
| 177 |
+
"44.self_attn.v_proj",
|
| 178 |
+
"29.self_attn.v_proj"
|
| 179 |
+
],
|
| 180 |
+
"target_parameters": null,
|
| 181 |
+
"task_type": null,
|
| 182 |
+
"trainable_token_indices": null,
|
| 183 |
+
"use_dora": false,
|
| 184 |
+
"use_qalora": false,
|
| 185 |
+
"use_rslora": false
|
| 186 |
+
}
|
12b/u_bottom-s0/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9325842fc069b24e9a81a07af1ea0338864791f85d84b6533275bf05736b9aac
|
| 3 |
+
size 1047628488
|
12b/u_bottom-s0/config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "google/gemma-3-12b-it",
|
| 3 |
+
"data": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma12/arms/TASKW-u_bottom-gemma12.jsonl",
|
| 4 |
+
"run": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma12/runs/u_bottom-s0",
|
| 5 |
+
"steps": 0,
|
| 6 |
+
"batch": 16,
|
| 7 |
+
"micro": 1,
|
| 8 |
+
"seq_len": 4096,
|
| 9 |
+
"lr": 0.0001,
|
| 10 |
+
"warmup": 10,
|
| 11 |
+
"lora_rank": 64,
|
| 12 |
+
"lora_targets": "all-linear",
|
| 13 |
+
"lora_targets_file": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/code/gemma3_12b_lora_targets.json",
|
| 14 |
+
"seed": 0,
|
| 15 |
+
"bos": true,
|
| 16 |
+
"no_merge": false,
|
| 17 |
+
"save_at": "",
|
| 18 |
+
"replay": "",
|
| 19 |
+
"replay_frac": 0.0,
|
| 20 |
+
"constant_lr": false,
|
| 21 |
+
"schedule": "hf-cosine",
|
| 22 |
+
"min_lr_frac": 0.0,
|
| 23 |
+
"warmup_ratio": 0.05,
|
| 24 |
+
"lora_dropout": 0.05,
|
| 25 |
+
"no_pack": true,
|
| 26 |
+
"chat_sft": true,
|
| 27 |
+
"data_frac": 1.0,
|
| 28 |
+
"base_adapter": "",
|
| 29 |
+
"ckpt_every": 0,
|
| 30 |
+
"kill_after": 0,
|
| 31 |
+
"method": "doc_lm_lora",
|
| 32 |
+
"world": 2
|
| 33 |
+
}
|
12b/u_bottom-s0/log.jsonl
ADDED
|
@@ -0,0 +1,312 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 1, "loss": 2.2796168327331543, "grad_norm": 4.400256633758545, "lr": 1.25e-05, "epoch": 1, "t": 6.529042482376099}
|
| 2 |
+
{"step": 2, "loss": 2.187624931335449, "grad_norm": 4.020505428314209, "lr": 1.8750000000000002e-05, "epoch": 1, "t": 5.776473045349121}
|
| 3 |
+
{"step": 3, "loss": 2.1392927169799805, "grad_norm": 3.6714041233062744, "lr": 2.5e-05, "epoch": 1, "t": 5.566001653671265}
|
| 4 |
+
{"step": 4, "loss": 1.9095730781555176, "grad_norm": 2.9393723011016846, "lr": 3.125e-05, "epoch": 1, "t": 5.125308513641357}
|
| 5 |
+
{"step": 5, "loss": 1.8427956104278564, "grad_norm": 2.1116926670074463, "lr": 3.7500000000000003e-05, "epoch": 1, "t": 6.0628790855407715}
|
| 6 |
+
{"step": 6, "loss": 1.7365288734436035, "grad_norm": 1.4905673265457153, "lr": 4.375e-05, "epoch": 1, "t": 4.792157888412476}
|
| 7 |
+
{"step": 7, "loss": 1.6273750066757202, "grad_norm": 1.0730162858963013, "lr": 5e-05, "epoch": 1, "t": 3.939953088760376}
|
| 8 |
+
{"step": 8, "loss": 1.5020382404327393, "grad_norm": 0.8699246644973755, "lr": 5.6250000000000005e-05, "epoch": 1, "t": 4.6516852378845215}
|
| 9 |
+
{"step": 9, "loss": 1.6671159267425537, "grad_norm": 0.9875266551971436, "lr": 6.25e-05, "epoch": 1, "t": 4.667775630950928}
|
| 10 |
+
{"step": 10, "loss": 1.7133204936981201, "grad_norm": 1.0434733629226685, "lr": 6.875e-05, "epoch": 1, "t": 5.036233186721802}
|
| 11 |
+
{"step": 11, "loss": 1.4195690155029297, "grad_norm": 0.9061106443405151, "lr": 7.500000000000001e-05, "epoch": 1, "t": 4.639129161834717}
|
| 12 |
+
{"step": 12, "loss": 1.4301880598068237, "grad_norm": 0.8332769274711609, "lr": 8.125000000000001e-05, "epoch": 1, "t": 4.400840520858765}
|
| 13 |
+
{"step": 13, "loss": 1.2335591316223145, "grad_norm": 0.6944305300712585, "lr": 8.75e-05, "epoch": 1, "t": 3.5247182846069336}
|
| 14 |
+
{"step": 14, "loss": 1.598285436630249, "grad_norm": 0.8040773868560791, "lr": 9.375e-05, "epoch": 1, "t": 3.1271800994873047}
|
| 15 |
+
{"step": 15, "loss": 1.4832850694656372, "grad_norm": 0.7591297626495361, "lr": 0.0001, "epoch": 1, "t": 3.4279708862304688}
|
| 16 |
+
{"step": 16, "loss": 1.376697301864624, "grad_norm": 0.735444962978363, "lr": 0.0001, "epoch": 1, "t": 3.4197046756744385}
|
| 17 |
+
{"step": 17, "loss": 1.3606016635894775, "grad_norm": 0.7660970091819763, "lr": 9.99971838728789e-05, "epoch": 1, "t": 3.7712576389312744}
|
| 18 |
+
{"step": 18, "loss": 1.3299849033355713, "grad_norm": 0.7133594155311584, "lr": 9.998873580873848e-05, "epoch": 1, "t": 3.6111791133880615}
|
| 19 |
+
{"step": 19, "loss": 1.2998201847076416, "grad_norm": 0.7617425918579102, "lr": 9.997465675921163e-05, "epoch": 1, "t": 4.082550764083862}
|
| 20 |
+
{"step": 20, "loss": 1.3217943906784058, "grad_norm": 0.6228795051574707, "lr": 9.995494831023409e-05, "epoch": 1, "t": 3.914626121520996}
|
| 21 |
+
{"step": 21, "loss": 1.262622356414795, "grad_norm": 0.6451112627983093, "lr": 9.992961268186573e-05, "epoch": 1, "t": 3.759744644165039}
|
| 22 |
+
{"step": 22, "loss": 1.3451486825942993, "grad_norm": 0.6689098477363586, "lr": 9.989865272804063e-05, "epoch": 1, "t": 3.9024102687835693}
|
| 23 |
+
{"step": 23, "loss": 1.1713173389434814, "grad_norm": 0.6071130633354187, "lr": 9.986207193624536e-05, "epoch": 1, "t": 3.8475289344787598}
|
| 24 |
+
{"step": 24, "loss": 1.3761532306671143, "grad_norm": 0.5865582227706909, "lr": 9.981987442712633e-05, "epoch": 1, "t": 3.609874725341797}
|
| 25 |
+
{"step": 25, "loss": 1.1579132080078125, "grad_norm": 0.5750421285629272, "lr": 9.977206495402554e-05, "epoch": 1, "t": 3.236706495285034}
|
| 26 |
+
{"step": 26, "loss": 1.3194642066955566, "grad_norm": 0.6225281953811646, "lr": 9.971864890244513e-05, "epoch": 1, "t": 3.1698687076568604}
|
| 27 |
+
{"step": 27, "loss": 1.3758137226104736, "grad_norm": 0.5653595924377441, "lr": 9.965963228944078e-05, "epoch": 1, "t": 3.1658356189727783}
|
| 28 |
+
{"step": 28, "loss": 1.3184517621994019, "grad_norm": 0.6315352916717529, "lr": 9.959502176294383e-05, "epoch": 1, "t": 2.774376630783081}
|
| 29 |
+
{"step": 29, "loss": 1.247741460800171, "grad_norm": 0.7167669534683228, "lr": 9.95248246010126e-05, "epoch": 1, "t": 2.7905900478363037}
|
| 30 |
+
{"step": 30, "loss": 1.3454320430755615, "grad_norm": 0.6151255965232849, "lr": 9.944904871101228e-05, "epoch": 1, "t": 2.8834762573242188}
|
| 31 |
+
{"step": 31, "loss": 1.2908504009246826, "grad_norm": 0.5934525728225708, "lr": 9.936770262872443e-05, "epoch": 1, "t": 2.8511104583740234}
|
| 32 |
+
{"step": 32, "loss": 1.246272087097168, "grad_norm": 0.5656528472900391, "lr": 9.928079551738543e-05, "epoch": 1, "t": 3.3971545696258545}
|
| 33 |
+
{"step": 33, "loss": 1.2167012691497803, "grad_norm": 0.5671595931053162, "lr": 9.918833716665419e-05, "epoch": 1, "t": 3.0128860473632812}
|
| 34 |
+
{"step": 34, "loss": 1.1612294912338257, "grad_norm": 0.6264058351516724, "lr": 9.909033799150946e-05, "epoch": 1, "t": 3.151306390762329}
|
| 35 |
+
{"step": 35, "loss": 1.1641662120819092, "grad_norm": 0.5991170406341553, "lr": 9.898680903107666e-05, "epoch": 1, "t": 2.8719534873962402}
|
| 36 |
+
{"step": 36, "loss": 1.2150335311889648, "grad_norm": 0.5848028063774109, "lr": 9.887776194738432e-05, "epoch": 1, "t": 3.593238592147827}
|
| 37 |
+
{"step": 37, "loss": 1.2703022956848145, "grad_norm": 0.6096404790878296, "lr": 9.876320902405042e-05, "epoch": 1, "t": 3.589287757873535}
|
| 38 |
+
{"step": 38, "loss": 1.2524504661560059, "grad_norm": 0.6689551472663879, "lr": 9.864316316489873e-05, "epoch": 1, "t": 2.939807891845703}
|
| 39 |
+
{"step": 39, "loss": 1.3208110332489014, "grad_norm": 0.6150893568992615, "lr": 9.851763789250525e-05, "epoch": 1, "t": 2.79067325592041}
|
| 40 |
+
{"step": 40, "loss": 1.2492326498031616, "grad_norm": 0.5812289118766785, "lr": 9.838664734667495e-05, "epoch": 1, "t": 3.183887243270874}
|
| 41 |
+
{"step": 41, "loss": 1.1764318943023682, "grad_norm": 0.6673151850700378, "lr": 9.825020628284896e-05, "epoch": 1, "t": 3.1451289653778076}
|
| 42 |
+
{"step": 42, "loss": 1.269309401512146, "grad_norm": 0.5575243234634399, "lr": 9.810833007044247e-05, "epoch": 1, "t": 3.308140277862549}
|
| 43 |
+
{"step": 43, "loss": 1.3327550888061523, "grad_norm": 0.5585793256759644, "lr": 9.796103469111351e-05, "epoch": 1, "t": 4.341813325881958}
|
| 44 |
+
{"step": 44, "loss": 1.217625379562378, "grad_norm": 0.5931463837623596, "lr": 9.780833673696254e-05, "epoch": 1, "t": 3.953521728515625}
|
| 45 |
+
{"step": 45, "loss": 1.2160354852676392, "grad_norm": 0.5486651659011841, "lr": 9.76502534086636e-05, "epoch": 1, "t": 3.840898275375366}
|
| 46 |
+
{"step": 46, "loss": 1.2350542545318604, "grad_norm": 0.5826579928398132, "lr": 9.74868025135266e-05, "epoch": 1, "t": 3.692005157470703}
|
| 47 |
+
{"step": 47, "loss": 1.251184344291687, "grad_norm": 0.577694296836853, "lr": 9.731800246349148e-05, "epoch": 1, "t": 3.7976977825164795}
|
| 48 |
+
{"step": 48, "loss": 1.2458150386810303, "grad_norm": 0.5890681743621826, "lr": 9.714387227305422e-05, "epoch": 1, "t": 3.81876802444458}
|
| 49 |
+
{"step": 49, "loss": 1.1998258829116821, "grad_norm": 0.5721865892410278, "lr": 9.696443155712486e-05, "epoch": 1, "t": 3.5375852584838867}
|
| 50 |
+
{"step": 50, "loss": 1.2159050703048706, "grad_norm": 0.5733393430709839, "lr": 9.67797005288181e-05, "epoch": 1, "t": 4.073696851730347}
|
| 51 |
+
{"step": 51, "loss": 1.3022195100784302, "grad_norm": 0.5709722638130188, "lr": 9.65896999971763e-05, "epoch": 1, "t": 4.027852773666382}
|
| 52 |
+
{"step": 52, "loss": 1.235236644744873, "grad_norm": 0.6509473919868469, "lr": 9.639445136482548e-05, "epoch": 1, "t": 3.913759708404541}
|
| 53 |
+
{"step": 53, "loss": 1.1816620826721191, "grad_norm": 0.6693188548088074, "lr": 9.619397662556435e-05, "epoch": 1, "t": 3.5523266792297363}
|
| 54 |
+
{"step": 54, "loss": 1.3099932670593262, "grad_norm": 0.6470484137535095, "lr": 9.598829836188694e-05, "epoch": 1, "t": 3.910655975341797}
|
| 55 |
+
{"step": 55, "loss": 1.2423217296600342, "grad_norm": 0.6475738286972046, "lr": 9.577743974243874e-05, "epoch": 1, "t": 3.8942806720733643}
|
| 56 |
+
{"step": 56, "loss": 1.1963086128234863, "grad_norm": 0.5847989320755005, "lr": 9.55614245194068e-05, "epoch": 1, "t": 3.4791030883789062}
|
| 57 |
+
{"step": 57, "loss": 1.2429912090301514, "grad_norm": 0.6662818193435669, "lr": 9.534027702584425e-05, "epoch": 1, "t": 3.1651899814605713}
|
| 58 |
+
{"step": 58, "loss": 1.1542389392852783, "grad_norm": 0.5899723768234253, "lr": 9.511402217292926e-05, "epoch": 1, "t": 3.2321722507476807}
|
| 59 |
+
{"step": 59, "loss": 1.2696019411087036, "grad_norm": 0.5878300666809082, "lr": 9.488268544715896e-05, "epoch": 1, "t": 2.989182233810425}
|
| 60 |
+
{"step": 60, "loss": 1.2108500003814697, "grad_norm": 0.5927729606628418, "lr": 9.464629290747842e-05, "epoch": 1, "t": 3.285382032394409}
|
| 61 |
+
{"step": 61, "loss": 1.4031319618225098, "grad_norm": 0.5806950926780701, "lr": 9.440487118234535e-05, "epoch": 1, "t": 3.1141865253448486}
|
| 62 |
+
{"step": 62, "loss": 1.2543126344680786, "grad_norm": 0.5452511310577393, "lr": 9.415844746673047e-05, "epoch": 1, "t": 2.9406721591949463}
|
| 63 |
+
{"step": 63, "loss": 1.2066779136657715, "grad_norm": 0.5408366918563843, "lr": 9.390704951905411e-05, "epoch": 1, "t": 2.9355382919311523}
|
| 64 |
+
{"step": 64, "loss": 1.2902932167053223, "grad_norm": 0.5582548379898071, "lr": 9.365070565805941e-05, "epoch": 1, "t": 3.316260814666748}
|
| 65 |
+
{"step": 65, "loss": 1.1782703399658203, "grad_norm": 0.5738767385482788, "lr": 9.338944475962237e-05, "epoch": 1, "t": 3.3220856189727783}
|
| 66 |
+
{"step": 66, "loss": 1.2352807521820068, "grad_norm": 0.6358559131622314, "lr": 9.312329625349902e-05, "epoch": 1, "t": 2.515019178390503}
|
| 67 |
+
{"step": 67, "loss": 1.3366938829421997, "grad_norm": 0.6683616042137146, "lr": 9.285229012001047e-05, "epoch": 1, "t": 3.324474573135376}
|
| 68 |
+
{"step": 68, "loss": 1.1387542486190796, "grad_norm": 0.5232688784599304, "lr": 9.257645688666556e-05, "epoch": 1, "t": 3.4642415046691895}
|
| 69 |
+
{"step": 69, "loss": 1.1257469654083252, "grad_norm": 0.5826013088226318, "lr": 9.22958276247223e-05, "epoch": 1, "t": 2.810372829437256}
|
| 70 |
+
{"step": 70, "loss": 1.4037599563598633, "grad_norm": 0.6210429668426514, "lr": 9.201043394568773e-05, "epoch": 1, "t": 3.4884438514709473}
|
| 71 |
+
{"step": 71, "loss": 1.0940579175949097, "grad_norm": 0.6129340529441833, "lr": 9.172030799775699e-05, "epoch": 1, "t": 2.799748420715332}
|
| 72 |
+
{"step": 72, "loss": 1.2496837377548218, "grad_norm": 0.5619972944259644, "lr": 9.142548246219212e-05, "epoch": 1, "t": 3.32836651802063}
|
| 73 |
+
{"step": 73, "loss": 1.0862205028533936, "grad_norm": 0.6886274218559265, "lr": 9.112599054964057e-05, "epoch": 1, "t": 2.5285918712615967}
|
| 74 |
+
{"step": 74, "loss": 1.1376888751983643, "grad_norm": 0.5930537581443787, "lr": 9.082186599639428e-05, "epoch": 1, "t": 3.0229763984680176}
|
| 75 |
+
{"step": 75, "loss": 1.1845810413360596, "grad_norm": 0.8033909201622009, "lr": 9.051314306058933e-05, "epoch": 1, "t": 2.9015369415283203}
|
| 76 |
+
{"step": 76, "loss": 1.1939804553985596, "grad_norm": 0.6111111044883728, "lr": 9.019985651834703e-05, "epoch": 1, "t": 2.897002696990967}
|
| 77 |
+
{"step": 77, "loss": 1.220609426498413, "grad_norm": 0.6012175679206848, "lr": 8.988204165985649e-05, "epoch": 1, "t": 3.34718656539917}
|
| 78 |
+
{"step": 78, "loss": 1.1667402982711792, "grad_norm": 0.594720721244812, "lr": 8.955973428539944e-05, "epoch": 1, "t": 2.8113768100738525}
|
| 79 |
+
{"step": 79, "loss": 1.2189438343048096, "grad_norm": 0.5567514300346375, "lr": 8.923297070131737e-05, "epoch": 1, "t": 2.933164596557617}
|
| 80 |
+
{"step": 80, "loss": 1.1402318477630615, "grad_norm": 0.6345027685165405, "lr": 8.890178771592199e-05, "epoch": 1, "t": 2.7314743995666504}
|
| 81 |
+
{"step": 81, "loss": 1.2136492729187012, "grad_norm": 0.6385586261749268, "lr": 8.856622263534875e-05, "epoch": 1, "t": 3.028803586959839}
|
| 82 |
+
{"step": 82, "loss": 1.2703094482421875, "grad_norm": 0.6952808499336243, "lr": 8.822631325935463e-05, "epoch": 1, "t": 2.77531099319458}
|
| 83 |
+
{"step": 83, "loss": 1.352795124053955, "grad_norm": 0.6440946459770203, "lr": 8.788209787706015e-05, "epoch": 1, "t": 3.168881416320801}
|
| 84 |
+
{"step": 84, "loss": 1.1485934257507324, "grad_norm": 0.650083065032959, "lr": 8.753361526263621e-05, "epoch": 1, "t": 2.8522775173187256}
|
| 85 |
+
{"step": 85, "loss": 0.9985901117324829, "grad_norm": 0.8387364149093628, "lr": 8.718090467093654e-05, "epoch": 1, "t": 3.4434592723846436}
|
| 86 |
+
{"step": 86, "loss": 1.0919952392578125, "grad_norm": 0.5590276122093201, "lr": 8.682400583307562e-05, "epoch": 1, "t": 3.630811929702759}
|
| 87 |
+
{"step": 87, "loss": 1.3228874206542969, "grad_norm": 0.5738188624382019, "lr": 8.646295895195333e-05, "epoch": 1, "t": 3.2759687900543213}
|
| 88 |
+
{"step": 88, "loss": 1.038428544998169, "grad_norm": 0.6042227745056152, "lr": 8.609780469772623e-05, "epoch": 1, "t": 3.297309398651123}
|
| 89 |
+
{"step": 89, "loss": 1.2378544807434082, "grad_norm": 0.5510876774787903, "lr": 8.572858420322627e-05, "epoch": 1, "t": 3.615286111831665}
|
| 90 |
+
{"step": 90, "loss": 1.279449224472046, "grad_norm": 0.5546838641166687, "lr": 8.535533905932738e-05, "epoch": 1, "t": 3.602992296218872}
|
| 91 |
+
{"step": 91, "loss": 1.1821234226226807, "grad_norm": 0.5775654911994934, "lr": 8.497811131026046e-05, "epoch": 1, "t": 3.2568318843841553}
|
| 92 |
+
{"step": 92, "loss": 1.191941499710083, "grad_norm": 0.5937525033950806, "lr": 8.459694344887732e-05, "epoch": 1, "t": 3.696549415588379}
|
| 93 |
+
{"step": 93, "loss": 1.1268746852874756, "grad_norm": 0.5636759996414185, "lr": 8.421187841186402e-05, "epoch": 1, "t": 3.7892062664031982}
|
| 94 |
+
{"step": 94, "loss": 1.3100491762161255, "grad_norm": 0.5433634519577026, "lr": 8.382295957490436e-05, "epoch": 1, "t": 4.484153747558594}
|
| 95 |
+
{"step": 95, "loss": 1.2875268459320068, "grad_norm": 0.557334840297699, "lr": 8.343023074779368e-05, "epoch": 1, "t": 4.36227011680603}
|
| 96 |
+
{"step": 96, "loss": 1.2306948900222778, "grad_norm": 0.6284522414207458, "lr": 8.303373616950408e-05, "epoch": 1, "t": 4.398929119110107}
|
| 97 |
+
{"step": 97, "loss": 1.2416801452636719, "grad_norm": 0.6157611608505249, "lr": 8.263352050320094e-05, "epoch": 1, "t": 4.373669624328613}
|
| 98 |
+
{"step": 98, "loss": 1.24025297164917, "grad_norm": 0.6044377088546753, "lr": 8.222962883121196e-05, "epoch": 1, "t": 4.606372594833374}
|
| 99 |
+
{"step": 99, "loss": 1.22691011428833, "grad_norm": 0.6472576856613159, "lr": 8.182210664994878e-05, "epoch": 1, "t": 4.3771491050720215}
|
| 100 |
+
{"step": 100, "loss": 1.0906577110290527, "grad_norm": 0.5523853898048401, "lr": 8.141099986478212e-05, "epoch": 1, "t": 3.805881977081299}
|
| 101 |
+
{"step": 101, "loss": 1.04519784450531, "grad_norm": 0.5623805522918701, "lr": 8.099635478487064e-05, "epoch": 1, "t": 3.2478997707366943}
|
| 102 |
+
{"step": 102, "loss": 1.1493608951568604, "grad_norm": 0.5591173768043518, "lr": 8.057821811794458e-05, "epoch": 1, "t": 4.073742389678955}
|
| 103 |
+
{"step": 103, "loss": 1.1927741765975952, "grad_norm": 0.6300081014633179, "lr": 8.015663696504422e-05, "epoch": 1, "t": 4.002060651779175}
|
| 104 |
+
{"step": 104, "loss": 1.1215739250183105, "grad_norm": 0.5657426714897156, "lr": 7.973165881521434e-05, "epoch": 1, "t": 3.805173635482788}
|
| 105 |
+
{"step": 105, "loss": 1.110884666442871, "grad_norm": 0.635918915271759, "lr": 7.930333154015466e-05, "epoch": 1, "t": 4.056977033615112}
|
| 106 |
+
{"step": 106, "loss": 1.2040023803710938, "grad_norm": 0.5650188326835632, "lr": 7.88717033888274e-05, "epoch": 1, "t": 3.434739351272583}
|
| 107 |
+
{"step": 107, "loss": 1.1819329261779785, "grad_norm": 0.6216996312141418, "lr": 7.843682298202235e-05, "epoch": 1, "t": 3.6214869022369385}
|
| 108 |
+
{"step": 108, "loss": 1.1873865127563477, "grad_norm": 0.5492762923240662, "lr": 7.799873930687978e-05, "epoch": 1, "t": 4.112046718597412}
|
| 109 |
+
{"step": 109, "loss": 1.2666281461715698, "grad_norm": 0.579659104347229, "lr": 7.755750171137246e-05, "epoch": 1, "t": 4.060079336166382}
|
| 110 |
+
{"step": 110, "loss": 1.2372691631317139, "grad_norm": 0.6457576155662537, "lr": 7.711315989874677e-05, "epoch": 1, "t": 3.6417388916015625}
|
| 111 |
+
{"step": 111, "loss": 1.2777879238128662, "grad_norm": 0.6187204718589783, "lr": 7.666576392192389e-05, "epoch": 1, "t": 4.148644924163818}
|
| 112 |
+
{"step": 112, "loss": 1.0683177709579468, "grad_norm": 0.6000900864601135, "lr": 7.621536417786159e-05, "epoch": 1, "t": 3.908047914505005}
|
| 113 |
+
{"step": 113, "loss": 1.2187702655792236, "grad_norm": 0.6108975410461426, "lr": 7.576201140187727e-05, "epoch": 1, "t": 3.8268630504608154}
|
| 114 |
+
{"step": 114, "loss": 1.203293800354004, "grad_norm": 0.6008256077766418, "lr": 7.530575666193283e-05, "epoch": 1, "t": 3.799168109893799}
|
| 115 |
+
{"step": 115, "loss": 1.1572589874267578, "grad_norm": 0.646173357963562, "lr": 7.484665135288213e-05, "epoch": 1, "t": 3.906832695007324}
|
| 116 |
+
{"step": 116, "loss": 1.107800006866455, "grad_norm": 0.6491538286209106, "lr": 7.438474719068173e-05, "epoch": 1, "t": 3.9282758235931396}
|
| 117 |
+
{"step": 117, "loss": 1.1002235412597656, "grad_norm": 0.7050867676734924, "lr": 7.392009620656513e-05, "epoch": 1, "t": 3.2994375228881836}
|
| 118 |
+
{"step": 118, "loss": 1.0847188234329224, "grad_norm": 0.675769567489624, "lr": 7.345275074118185e-05, "epoch": 1, "t": 3.962752103805542}
|
| 119 |
+
{"step": 119, "loss": 1.1019296646118164, "grad_norm": 0.5060476660728455, "lr": 7.298276343870151e-05, "epoch": 1, "t": 4.020556926727295}
|
| 120 |
+
{"step": 120, "loss": 1.1388767957687378, "grad_norm": 0.5517978072166443, "lr": 7.251018724088367e-05, "epoch": 1, "t": 4.11132550239563}
|
| 121 |
+
{"step": 121, "loss": 1.1106231212615967, "grad_norm": 0.6458324193954468, "lr": 7.203507538111423e-05, "epoch": 1, "t": 3.6184656620025635}
|
| 122 |
+
{"step": 122, "loss": 1.1430410146713257, "grad_norm": 0.6245289444923401, "lr": 7.155748137840892e-05, "epoch": 1, "t": 3.472459077835083}
|
| 123 |
+
{"step": 123, "loss": 1.194145917892456, "grad_norm": 0.5574867725372314, "lr": 7.107745903138472e-05, "epoch": 1, "t": 3.8255302906036377}
|
| 124 |
+
{"step": 124, "loss": 1.293508529663086, "grad_norm": 0.6076470017433167, "lr": 7.059506241219965e-05, "epoch": 1, "t": 3.902667760848999}
|
| 125 |
+
{"step": 125, "loss": 1.1743580102920532, "grad_norm": 0.5984256267547607, "lr": 7.011034586046176e-05, "epoch": 1, "t": 3.908402681350708}
|
| 126 |
+
{"step": 126, "loss": 1.121490240097046, "grad_norm": 0.5764150619506836, "lr": 6.962336397710819e-05, "epoch": 1, "t": 3.927380084991455}
|
| 127 |
+
{"step": 127, "loss": 1.1504158973693848, "grad_norm": 0.6502886414527893, "lr": 6.91341716182545e-05, "epoch": 1, "t": 3.7148592472076416}
|
| 128 |
+
{"step": 128, "loss": 1.1956489086151123, "grad_norm": 0.6529905796051025, "lr": 6.864282388901544e-05, "epoch": 1, "t": 3.876858949661255}
|
| 129 |
+
{"step": 129, "loss": 1.167960286140442, "grad_norm": 0.5582114458084106, "lr": 6.814937613729766e-05, "epoch": 1, "t": 3.877762794494629}
|
| 130 |
+
{"step": 130, "loss": 1.165308952331543, "grad_norm": 0.6209343671798706, "lr": 6.765388394756504e-05, "epoch": 1, "t": 3.7977523803710938}
|
| 131 |
+
{"step": 131, "loss": 1.0924253463745117, "grad_norm": 0.5518262982368469, "lr": 6.715640313457733e-05, "epoch": 1, "t": 3.5653810501098633}
|
| 132 |
+
{"step": 132, "loss": 1.139561653137207, "grad_norm": 0.5491393208503723, "lr": 6.665698973710288e-05, "epoch": 1, "t": 3.6677472591400146}
|
| 133 |
+
{"step": 133, "loss": 1.0036296844482422, "grad_norm": 0.5746570229530334, "lr": 6.615570001160626e-05, "epoch": 1, "t": 3.7977983951568604}
|
| 134 |
+
{"step": 134, "loss": 1.247854471206665, "grad_norm": 0.576014518737793, "lr": 6.565259042591113e-05, "epoch": 1, "t": 4.056816101074219}
|
| 135 |
+
{"step": 135, "loss": 1.1368107795715332, "grad_norm": 0.6416566371917725, "lr": 6.514771765283942e-05, "epoch": 1, "t": 2.689995765686035}
|
| 136 |
+
{"step": 136, "loss": 1.1421349048614502, "grad_norm": 0.7293333411216736, "lr": 6.464113856382752e-05, "epoch": 1, "t": 2.9442124366760254}
|
| 137 |
+
{"step": 137, "loss": 1.1195770502090454, "grad_norm": 0.5645886659622192, "lr": 6.413291022251989e-05, "epoch": 1, "t": 3.0367612838745117}
|
| 138 |
+
{"step": 138, "loss": 1.2760584354400635, "grad_norm": 0.5907130837440491, "lr": 6.362308987834115e-05, "epoch": 1, "t": 3.262828826904297}
|
| 139 |
+
{"step": 139, "loss": 1.0437781810760498, "grad_norm": 0.6388912200927734, "lr": 6.311173496004723e-05, "epoch": 1, "t": 2.9264111518859863}
|
| 140 |
+
{"step": 140, "loss": 1.2465133666992188, "grad_norm": 0.5667279362678528, "lr": 6.259890306925627e-05, "epoch": 1, "t": 3.1224493980407715}
|
| 141 |
+
{"step": 141, "loss": 1.1778254508972168, "grad_norm": 0.7854788303375244, "lr": 6.208465197396013e-05, "epoch": 1, "t": 2.786785364151001}
|
| 142 |
+
{"step": 142, "loss": 1.0598783493041992, "grad_norm": 0.6885096430778503, "lr": 6.156903960201709e-05, "epoch": 1, "t": 2.88132643699646}
|
| 143 |
+
{"step": 143, "loss": 1.193947196006775, "grad_norm": 0.6000897288322449, "lr": 6.105212403462651e-05, "epoch": 1, "t": 3.3030450344085693}
|
| 144 |
+
{"step": 144, "loss": 1.2082865238189697, "grad_norm": 0.6318554878234863, "lr": 6.0533963499786314e-05, "epoch": 1, "t": 2.8742475509643555}
|
| 145 |
+
{"step": 145, "loss": 1.200656533241272, "grad_norm": 0.5822675824165344, "lr": 6.001461636573397e-05, "epoch": 1, "t": 2.9299895763397217}
|
| 146 |
+
{"step": 146, "loss": 1.1954030990600586, "grad_norm": 0.5650563836097717, "lr": 5.949414113437142e-05, "epoch": 1, "t": 2.949399709701538}
|
| 147 |
+
{"step": 147, "loss": 1.1622741222381592, "grad_norm": 0.568586528301239, "lr": 5.897259643467527e-05, "epoch": 1, "t": 2.968006134033203}
|
| 148 |
+
{"step": 148, "loss": 1.100928544998169, "grad_norm": 0.5402028560638428, "lr": 5.8450041016092464e-05, "epoch": 1, "t": 3.0402121543884277}
|
| 149 |
+
{"step": 149, "loss": 1.1039650440216064, "grad_norm": 0.6397644877433777, "lr": 5.792653374192245e-05, "epoch": 1, "t": 2.7922537326812744}
|
| 150 |
+
{"step": 150, "loss": 1.242647409439087, "grad_norm": 0.5479434728622437, "lr": 5.7402133582686576e-05, "epoch": 1, "t": 3.6813130378723145}
|
| 151 |
+
{"step": 151, "loss": 1.2494502067565918, "grad_norm": 0.6544719934463501, "lr": 5.6876899609485256e-05, "epoch": 1, "t": 2.8223929405212402}
|
| 152 |
+
{"step": 152, "loss": 1.076706886291504, "grad_norm": 0.6050904989242554, "lr": 5.6350890987343944e-05, "epoch": 1, "t": 3.354572296142578}
|
| 153 |
+
{"step": 153, "loss": 1.2764182090759277, "grad_norm": 0.5722780227661133, "lr": 5.582416696854853e-05, "epoch": 1, "t": 3.2717432975769043}
|
| 154 |
+
{"step": 154, "loss": 1.196759581565857, "grad_norm": 0.6652747988700867, "lr": 5.5296786885970805e-05, "epoch": 1, "t": 2.904327154159546}
|
| 155 |
+
{"step": 155, "loss": 1.1735427379608154, "grad_norm": 0.6104219555854797, "lr": 5.476881014638491e-05, "epoch": 1, "t": 2.8878519535064697}
|
| 156 |
+
{"step": 156, "loss": 1.2194571495056152, "grad_norm": 0.590528666973114, "lr": 5.4240296223775465e-05, "epoch": 1, "t": 3.2303807735443115}
|
| 157 |
+
{"step": 157, "loss": 1.159003496170044, "grad_norm": 0.5662127137184143, "lr": 5.3711304652638126e-05, "epoch": 1, "t": 3.216623067855835}
|
| 158 |
+
{"step": 158, "loss": 1.0291956663131714, "grad_norm": 0.653940737247467, "lr": 5.318189502127332e-05, "epoch": 1, "t": 2.6618824005126953}
|
| 159 |
+
{"step": 159, "loss": 1.1841682195663452, "grad_norm": 0.5722079277038574, "lr": 5.265212696507387e-05, "epoch": 1, "t": 3.252058744430542}
|
| 160 |
+
{"step": 160, "loss": 1.2371938228607178, "grad_norm": 0.5706830620765686, "lr": 5.212206015980742e-05, "epoch": 1, "t": 3.335721015930176}
|
| 161 |
+
{"step": 161, "loss": 1.096413254737854, "grad_norm": 0.5922949910163879, "lr": 5.159175431489424e-05, "epoch": 1, "t": 2.917912721633911}
|
| 162 |
+
{"step": 162, "loss": 1.1713407039642334, "grad_norm": 0.6230096817016602, "lr": 5.1061269166681183e-05, "epoch": 1, "t": 3.006758689880371}
|
| 163 |
+
{"step": 163, "loss": 0.9827152490615845, "grad_norm": 0.5703536868095398, "lr": 5.053066447171282e-05, "epoch": 1, "t": 3.10632586479187}
|
| 164 |
+
{"step": 164, "loss": 1.1528689861297607, "grad_norm": 0.5780932307243347, "lr": 5e-05, "epoch": 1, "t": 3.277851104736328}
|
| 165 |
+
{"step": 165, "loss": 1.0204691886901855, "grad_norm": 0.5498275756835938, "lr": 4.94693355282872e-05, "epoch": 1, "t": 2.549923896789551}
|
| 166 |
+
{"step": 166, "loss": 1.2198524475097656, "grad_norm": 0.5634288787841797, "lr": 4.893873083331882e-05, "epoch": 1, "t": 3.530629873275757}
|
| 167 |
+
{"step": 167, "loss": 1.1341540813446045, "grad_norm": 0.5160549283027649, "lr": 4.840824568510579e-05, "epoch": 1, "t": 3.4451637268066406}
|
| 168 |
+
{"step": 168, "loss": 1.189644455909729, "grad_norm": 0.6211481094360352, "lr": 4.78779398401926e-05, "epoch": 1, "t": 3.210346221923828}
|
| 169 |
+
{"step": 169, "loss": 1.131641149520874, "grad_norm": 0.6778606176376343, "lr": 4.734787303492615e-05, "epoch": 1, "t": 3.1198132038116455}
|
| 170 |
+
{"step": 170, "loss": 1.2129818201065063, "grad_norm": 0.6030727028846741, "lr": 4.6818104978726685e-05, "epoch": 1, "t": 3.0022897720336914}
|
| 171 |
+
{"step": 171, "loss": 1.1575881242752075, "grad_norm": 0.587956964969635, "lr": 4.628869534736187e-05, "epoch": 1, "t": 3.127443552017212}
|
| 172 |
+
{"step": 172, "loss": 1.081390619277954, "grad_norm": 0.699698805809021, "lr": 4.575970377622456e-05, "epoch": 1, "t": 2.531400203704834}
|
| 173 |
+
{"step": 173, "loss": 1.1386687755584717, "grad_norm": 0.5645583868026733, "lr": 4.52311898536151e-05, "epoch": 1, "t": 2.968369245529175}
|
| 174 |
+
{"step": 174, "loss": 1.152360200881958, "grad_norm": 0.5911768078804016, "lr": 4.47032131140292e-05, "epoch": 1, "t": 2.897966146469116}
|
| 175 |
+
{"step": 175, "loss": 1.1680355072021484, "grad_norm": 0.6221476197242737, "lr": 4.4175833031451473e-05, "epoch": 1, "t": 2.7802441120147705}
|
| 176 |
+
{"step": 176, "loss": 1.0546282529830933, "grad_norm": 0.5288382768630981, "lr": 4.364910901265606e-05, "epoch": 1, "t": 3.2952880859375}
|
| 177 |
+
{"step": 177, "loss": 1.1015164852142334, "grad_norm": 0.5779777765274048, "lr": 4.3123100390514756e-05, "epoch": 1, "t": 3.013418674468994}
|
| 178 |
+
{"step": 178, "loss": 1.0744432210922241, "grad_norm": 0.7750630378723145, "lr": 4.2597866417313436e-05, "epoch": 1, "t": 2.702049732208252}
|
| 179 |
+
{"step": 179, "loss": 1.1876572370529175, "grad_norm": 0.5808477997779846, "lr": 4.207346625807756e-05, "epoch": 1, "t": 2.766916036605835}
|
| 180 |
+
{"step": 180, "loss": 1.1583235263824463, "grad_norm": 0.5504623651504517, "lr": 4.1549958983907555e-05, "epoch": 1, "t": 3.5158584117889404}
|
| 181 |
+
{"step": 181, "loss": 1.1774816513061523, "grad_norm": 0.6100735664367676, "lr": 4.102740356532473e-05, "epoch": 1, "t": 3.2605581283569336}
|
| 182 |
+
{"step": 182, "loss": 1.1724961996078491, "grad_norm": 0.5525132417678833, "lr": 4.050585886562858e-05, "epoch": 1, "t": 2.8990583419799805}
|
| 183 |
+
{"step": 183, "loss": 1.1977407932281494, "grad_norm": 0.5886827707290649, "lr": 3.998538363426605e-05, "epoch": 1, "t": 3.36639666557312}
|
| 184 |
+
{"step": 184, "loss": 1.2504453659057617, "grad_norm": 0.5493999123573303, "lr": 3.94660365002137e-05, "epoch": 1, "t": 3.140617847442627}
|
| 185 |
+
{"step": 185, "loss": 1.065008521080017, "grad_norm": 0.5971235036849976, "lr": 3.894787596537352e-05, "epoch": 1, "t": 2.8115322589874268}
|
| 186 |
+
{"step": 186, "loss": 1.132164478302002, "grad_norm": 0.6075999736785889, "lr": 3.843096039798293e-05, "epoch": 1, "t": 2.861053943634033}
|
| 187 |
+
{"step": 187, "loss": 1.2062307596206665, "grad_norm": 0.6238131523132324, "lr": 3.791534802603988e-05, "epoch": 1, "t": 3.048740863800049}
|
| 188 |
+
{"step": 188, "loss": 1.0297080278396606, "grad_norm": 0.6358351707458496, "lr": 3.740109693074375e-05, "epoch": 1, "t": 2.929049253463745}
|
| 189 |
+
{"step": 189, "loss": 1.2059190273284912, "grad_norm": 0.5411284565925598, "lr": 3.68882650399528e-05, "epoch": 1, "t": 3.4467530250549316}
|
| 190 |
+
{"step": 190, "loss": 1.3108404874801636, "grad_norm": 0.6114359498023987, "lr": 3.637691012165886e-05, "epoch": 1, "t": 3.0853779315948486}
|
| 191 |
+
{"step": 191, "loss": 0.9888194799423218, "grad_norm": 0.6403577327728271, "lr": 3.586708977748012e-05, "epoch": 1, "t": 2.9144341945648193}
|
| 192 |
+
{"step": 192, "loss": 1.0799251794815063, "grad_norm": 0.590144693851471, "lr": 3.5358861436172485e-05, "epoch": 1, "t": 2.9566233158111572}
|
| 193 |
+
{"step": 193, "loss": 1.0908485651016235, "grad_norm": 0.5800775289535522, "lr": 3.485228234716058e-05, "epoch": 1, "t": 3.0918526649475098}
|
| 194 |
+
{"step": 194, "loss": 1.1255537271499634, "grad_norm": 0.5407876372337341, "lr": 3.434740957408889e-05, "epoch": 1, "t": 3.088022470474243}
|
| 195 |
+
{"step": 195, "loss": 0.9668008089065552, "grad_norm": 0.5884929895401001, "lr": 3.3844299988393755e-05, "epoch": 1, "t": 3.0009520053863525}
|
| 196 |
+
{"step": 196, "loss": 1.216160535812378, "grad_norm": 0.5410187840461731, "lr": 3.334301026289712e-05, "epoch": 1, "t": 3.3178014755249023}
|
| 197 |
+
{"step": 197, "loss": 1.1512436866760254, "grad_norm": 0.5714312791824341, "lr": 3.284359686542269e-05, "epoch": 1, "t": 3.00972580909729}
|
| 198 |
+
{"step": 198, "loss": 1.0415838956832886, "grad_norm": 0.6088261008262634, "lr": 3.234611605243496e-05, "epoch": 1, "t": 2.928483724594116}
|
| 199 |
+
{"step": 199, "loss": 1.0486640930175781, "grad_norm": 0.5961528420448303, "lr": 3.1850623862702344e-05, "epoch": 1, "t": 2.5869357585906982}
|
| 200 |
+
{"step": 200, "loss": 1.2178640365600586, "grad_norm": 0.614139974117279, "lr": 3.135717611098458e-05, "epoch": 1, "t": 3.117884397506714}
|
| 201 |
+
{"step": 201, "loss": 1.1999826431274414, "grad_norm": 0.6177778244018555, "lr": 3.086582838174551e-05, "epoch": 1, "t": 2.9931321144104004}
|
| 202 |
+
{"step": 202, "loss": 1.14932119846344, "grad_norm": 0.5659716725349426, "lr": 3.0376636022891812e-05, "epoch": 1, "t": 3.005204439163208}
|
| 203 |
+
{"step": 203, "loss": 1.0424413681030273, "grad_norm": 0.5663148760795593, "lr": 2.9889654139538246e-05, "epoch": 1, "t": 2.9001808166503906}
|
| 204 |
+
{"step": 204, "loss": 1.1409486532211304, "grad_norm": 0.6158498525619507, "lr": 2.9404937587800375e-05, "epoch": 1, "t": 3.241957664489746}
|
| 205 |
+
{"step": 205, "loss": 1.1239773035049438, "grad_norm": 0.6024658679962158, "lr": 2.8922540968615286e-05, "epoch": 1, "t": 3.4749953746795654}
|
| 206 |
+
{"step": 206, "loss": 1.1450831890106201, "grad_norm": 0.592876672744751, "lr": 2.8442518621591086e-05, "epoch": 1, "t": 2.6182096004486084}
|
| 207 |
+
{"step": 207, "loss": 1.1319626569747925, "grad_norm": 0.5627796053886414, "lr": 2.7964924618885778e-05, "epoch": 1, "t": 3.416124105453491}
|
| 208 |
+
{"step": 208, "loss": 1.1606276035308838, "grad_norm": 0.6031860113143921, "lr": 2.748981275911633e-05, "epoch": 1, "t": 3.349666118621826}
|
| 209 |
+
{"step": 209, "loss": 1.188281774520874, "grad_norm": 0.6572056412696838, "lr": 2.701723656129851e-05, "epoch": 1, "t": 2.6094274520874023}
|
| 210 |
+
{"step": 210, "loss": 1.0975607633590698, "grad_norm": 0.5507838129997253, "lr": 2.6547249258818164e-05, "epoch": 1, "t": 3.1348466873168945}
|
| 211 |
+
{"step": 211, "loss": 1.14078688621521, "grad_norm": 0.5925769209861755, "lr": 2.607990379343489e-05, "epoch": 1, "t": 2.9667868614196777}
|
| 212 |
+
{"step": 212, "loss": 1.0285526514053345, "grad_norm": 0.624167263507843, "lr": 2.5615252809318284e-05, "epoch": 1, "t": 3.144674777984619}
|
| 213 |
+
{"step": 213, "loss": 1.2753043174743652, "grad_norm": 0.5767782330513, "lr": 2.5153348647117857e-05, "epoch": 1, "t": 3.520312547683716}
|
| 214 |
+
{"step": 214, "loss": 1.2057697772979736, "grad_norm": 0.6363111138343811, "lr": 2.469424333806718e-05, "epoch": 1, "t": 3.1118662357330322}
|
| 215 |
+
{"step": 215, "loss": 1.075385332107544, "grad_norm": 0.5819528102874756, "lr": 2.4237988598122752e-05, "epoch": 1, "t": 2.9472646713256836}
|
| 216 |
+
{"step": 216, "loss": 1.0992224216461182, "grad_norm": 0.6043964624404907, "lr": 2.3784635822138424e-05, "epoch": 1, "t": 3.056764841079712}
|
| 217 |
+
{"step": 217, "loss": 1.151720404624939, "grad_norm": 0.5544857978820801, "lr": 2.333423607807613e-05, "epoch": 1, "t": 3.764554023742676}
|
| 218 |
+
{"step": 218, "loss": 1.0833635330200195, "grad_norm": 0.634046733379364, "lr": 2.288684010125325e-05, "epoch": 1, "t": 3.2840735912323}
|
| 219 |
+
{"step": 219, "loss": 1.1649768352508545, "grad_norm": 0.597759485244751, "lr": 2.2442498288627556e-05, "epoch": 1, "t": 2.8854196071624756}
|
| 220 |
+
{"step": 220, "loss": 1.0214139223098755, "grad_norm": 0.9099845886230469, "lr": 2.2001260693120233e-05, "epoch": 1, "t": 2.863187313079834}
|
| 221 |
+
{"step": 221, "loss": 1.1014854907989502, "grad_norm": 0.5698481202125549, "lr": 2.156317701797766e-05, "epoch": 1, "t": 3.1249442100524902}
|
| 222 |
+
{"step": 222, "loss": 1.1189522743225098, "grad_norm": 0.596674382686615, "lr": 2.1128296611172593e-05, "epoch": 1, "t": 2.7330100536346436}
|
| 223 |
+
{"step": 223, "loss": 1.1307971477508545, "grad_norm": 0.6788355112075806, "lr": 2.0696668459845355e-05, "epoch": 1, "t": 2.805807113647461}
|
| 224 |
+
{"step": 224, "loss": 1.2728497982025146, "grad_norm": 0.6415734887123108, "lr": 2.026834118478567e-05, "epoch": 1, "t": 2.9238686561584473}
|
| 225 |
+
{"step": 225, "loss": 1.0751233100891113, "grad_norm": 0.5854493975639343, "lr": 1.98433630349558e-05, "epoch": 1, "t": 3.3074238300323486}
|
| 226 |
+
{"step": 226, "loss": 1.020078420639038, "grad_norm": 0.6070003509521484, "lr": 1.9421781882055444e-05, "epoch": 1, "t": 2.811638355255127}
|
| 227 |
+
{"step": 227, "loss": 1.1844269037246704, "grad_norm": 0.6074785590171814, "lr": 1.9003645215129355e-05, "epoch": 1, "t": 2.951449155807495}
|
| 228 |
+
{"step": 228, "loss": 1.2330278158187866, "grad_norm": 0.6329712271690369, "lr": 1.858900013521788e-05, "epoch": 1, "t": 2.9545321464538574}
|
| 229 |
+
{"step": 229, "loss": 1.0916948318481445, "grad_norm": 0.5630735754966736, "lr": 1.817789335005121e-05, "epoch": 1, "t": 3.1227049827575684}
|
| 230 |
+
{"step": 230, "loss": 1.1919686794281006, "grad_norm": 0.5632869005203247, "lr": 1.777037116878804e-05, "epoch": 1, "t": 3.178375482559204}
|
| 231 |
+
{"step": 231, "loss": 1.1420263051986694, "grad_norm": 0.6705955266952515, "lr": 1.7366479496799077e-05, "epoch": 1, "t": 3.3353352546691895}
|
| 232 |
+
{"step": 232, "loss": 1.12916898727417, "grad_norm": 0.5830580592155457, "lr": 1.6966263830495936e-05, "epoch": 1, "t": 2.951209545135498}
|
| 233 |
+
{"step": 233, "loss": 1.100843906402588, "grad_norm": 0.5498954653739929, "lr": 1.656976925220633e-05, "epoch": 1, "t": 3.590996265411377}
|
| 234 |
+
{"step": 234, "loss": 1.1802067756652832, "grad_norm": 0.6611465811729431, "lr": 1.6177040425095662e-05, "epoch": 1, "t": 3.278940439224243}
|
| 235 |
+
{"step": 235, "loss": 1.2227891683578491, "grad_norm": 0.5920096039772034, "lr": 1.5788121588135975e-05, "epoch": 1, "t": 3.1620593070983887}
|
| 236 |
+
{"step": 236, "loss": 1.1182591915130615, "grad_norm": 0.5908201336860657, "lr": 1.5403056551122697e-05, "epoch": 1, "t": 3.068516731262207}
|
| 237 |
+
{"step": 237, "loss": 1.1879627704620361, "grad_norm": 0.5485942959785461, "lr": 1.5021888689739549e-05, "epoch": 1, "t": 3.147620439529419}
|
| 238 |
+
{"step": 238, "loss": 1.178399920463562, "grad_norm": 0.566916823387146, "lr": 1.4644660940672627e-05, "epoch": 1, "t": 3.0227084159851074}
|
| 239 |
+
{"step": 239, "loss": 1.1291710138320923, "grad_norm": 0.5659102201461792, "lr": 1.427141579677374e-05, "epoch": 1, "t": 3.0930042266845703}
|
| 240 |
+
{"step": 240, "loss": 1.123749017715454, "grad_norm": 0.5526357889175415, "lr": 1.3902195302273779e-05, "epoch": 1, "t": 4.072468280792236}
|
| 241 |
+
{"step": 241, "loss": 1.1211729049682617, "grad_norm": 0.5826315879821777, "lr": 1.3537041048046695e-05, "epoch": 1, "t": 4.41733980178833}
|
| 242 |
+
{"step": 242, "loss": 1.1045269966125488, "grad_norm": 0.6689655184745789, "lr": 1.3175994166924394e-05, "epoch": 1, "t": 4.44432806968689}
|
| 243 |
+
{"step": 243, "loss": 1.0280206203460693, "grad_norm": 0.5824991464614868, "lr": 1.2819095329063469e-05, "epoch": 1, "t": 4.081759452819824}
|
| 244 |
+
{"step": 244, "loss": 1.1183559894561768, "grad_norm": 0.7170051336288452, "lr": 1.246638473736378e-05, "epoch": 1, "t": 2.734480857849121}
|
| 245 |
+
{"step": 245, "loss": 1.1584432125091553, "grad_norm": 0.5693469047546387, "lr": 1.2117902122939861e-05, "epoch": 1, "t": 3.2850148677825928}
|
| 246 |
+
{"step": 246, "loss": 1.192258596420288, "grad_norm": 0.594908595085144, "lr": 1.1773686740645384e-05, "epoch": 1, "t": 3.2409627437591553}
|
| 247 |
+
{"step": 247, "loss": 1.0961769819259644, "grad_norm": 0.6132506132125854, "lr": 1.1433777364651271e-05, "epoch": 1, "t": 2.6660830974578857}
|
| 248 |
+
{"step": 248, "loss": 1.0398869514465332, "grad_norm": 0.6917411088943481, "lr": 1.1098212284078036e-05, "epoch": 1, "t": 2.73703670501709}
|
| 249 |
+
{"step": 249, "loss": 1.105663537979126, "grad_norm": 0.5996711850166321, "lr": 1.076702929868264e-05, "epoch": 1, "t": 3.0036585330963135}
|
| 250 |
+
{"step": 250, "loss": 1.0799379348754883, "grad_norm": 0.5624319911003113, "lr": 1.0440265714600572e-05, "epoch": 1, "t": 2.9499378204345703}
|
| 251 |
+
{"step": 251, "loss": 1.1669678688049316, "grad_norm": 0.5754649639129639, "lr": 1.0117958340143507e-05, "epoch": 1, "t": 3.080819845199585}
|
| 252 |
+
{"step": 252, "loss": 1.1613426208496094, "grad_norm": 0.5716949701309204, "lr": 9.800143481652979e-06, "epoch": 1, "t": 3.149203300476074}
|
| 253 |
+
{"step": 253, "loss": 1.1550145149230957, "grad_norm": 0.5361490249633789, "lr": 9.48685693941067e-06, "epoch": 1, "t": 3.073615789413452}
|
| 254 |
+
{"step": 254, "loss": 1.189213514328003, "grad_norm": 0.5783571600914001, "lr": 9.17813400360572e-06, "epoch": 1, "t": 3.475926160812378}
|
| 255 |
+
{"step": 255, "loss": 1.1734529733657837, "grad_norm": 0.5909482836723328, "lr": 8.874009450359427e-06, "epoch": 1, "t": 2.908905267715454}
|
| 256 |
+
{"step": 256, "loss": 1.1405507326126099, "grad_norm": 0.5495503544807434, "lr": 8.574517537807897e-06, "epoch": 1, "t": 3.0931739807128906}
|
| 257 |
+
{"step": 257, "loss": 1.0228874683380127, "grad_norm": 0.6029623746871948, "lr": 8.279692002243027e-06, "epoch": 1, "t": 2.623265027999878}
|
| 258 |
+
{"step": 258, "loss": 1.1184940338134766, "grad_norm": 0.5476334691047668, "lr": 7.989566054312287e-06, "epoch": 1, "t": 3.156865358352661}
|
| 259 |
+
{"step": 259, "loss": 0.9762853980064392, "grad_norm": 0.5764957070350647, "lr": 7.704172375277691e-06, "epoch": 1, "t": 3.138101100921631}
|
| 260 |
+
{"step": 260, "loss": 1.1840956211090088, "grad_norm": 0.6296178698539734, "lr": 7.423543113334436e-06, "epoch": 1, "t": 3.011345624923706}
|
| 261 |
+
{"step": 261, "loss": 1.0912525653839111, "grad_norm": 0.5177234411239624, "lr": 7.14770987998954e-06, "epoch": 1, "t": 3.1572554111480713}
|
| 262 |
+
{"step": 262, "loss": 1.2317930459976196, "grad_norm": 0.550730288028717, "lr": 6.876703746500984e-06, "epoch": 1, "t": 3.1815311908721924}
|
| 263 |
+
{"step": 263, "loss": 1.1975363492965698, "grad_norm": 0.6020235419273376, "lr": 6.610555240377652e-06, "epoch": 1, "t": 2.9622108936309814}
|
| 264 |
+
{"step": 264, "loss": 1.1572779417037964, "grad_norm": 0.648000955581665, "lr": 6.349294341940593e-06, "epoch": 1, "t": 2.724074363708496}
|
| 265 |
+
{"step": 265, "loss": 1.2379295825958252, "grad_norm": 0.6012561321258545, "lr": 6.092950480945897e-06, "epoch": 1, "t": 2.9164509773254395}
|
| 266 |
+
{"step": 266, "loss": 1.0732264518737793, "grad_norm": 0.6169740557670593, "lr": 5.841552533269534e-06, "epoch": 1, "t": 3.3635013103485107}
|
| 267 |
+
{"step": 267, "loss": 1.2429931163787842, "grad_norm": 0.5938675403594971, "lr": 5.595128817654638e-06, "epoch": 1, "t": 3.8237826824188232}
|
| 268 |
+
{"step": 268, "loss": 1.1276895999908447, "grad_norm": 0.552442729473114, "lr": 5.353707092521582e-06, "epoch": 1, "t": 3.568659782409668}
|
| 269 |
+
{"step": 269, "loss": 1.1378519535064697, "grad_norm": 0.6179390549659729, "lr": 5.117314552841052e-06, "epoch": 1, "t": 3.76232647895813}
|
| 270 |
+
{"step": 270, "loss": 1.1695225238800049, "grad_norm": 0.6306130886077881, "lr": 4.885977827070748e-06, "epoch": 1, "t": 3.8953850269317627}
|
| 271 |
+
{"step": 271, "loss": 1.1091785430908203, "grad_norm": 0.5891368985176086, "lr": 4.659722974155767e-06, "epoch": 1, "t": 3.540498971939087}
|
| 272 |
+
{"step": 272, "loss": 1.145094394683838, "grad_norm": 0.729819655418396, "lr": 4.43857548059321e-06, "epoch": 1, "t": 4.277384281158447}
|
| 273 |
+
{"step": 273, "loss": 1.1218918561935425, "grad_norm": 0.5778434872627258, "lr": 4.2225602575612755e-06, "epoch": 1, "t": 3.9088072776794434}
|
| 274 |
+
{"step": 274, "loss": 1.1688685417175293, "grad_norm": 0.6175435781478882, "lr": 4.011701638113063e-06, "epoch": 1, "t": 3.6113033294677734}
|
| 275 |
+
{"step": 275, "loss": 0.9873802661895752, "grad_norm": 0.5800295472145081, "lr": 3.8060233744356633e-06, "epoch": 1, "t": 3.9763290882110596}
|
| 276 |
+
{"step": 276, "loss": 1.1795215606689453, "grad_norm": 0.6106323599815369, "lr": 3.605548635174533e-06, "epoch": 1, "t": 4.0507025718688965}
|
| 277 |
+
{"step": 277, "loss": 1.121046781539917, "grad_norm": 0.5597973465919495, "lr": 3.410300002823691e-06, "epoch": 1, "t": 3.7010719776153564}
|
| 278 |
+
{"step": 278, "loss": 1.0764379501342773, "grad_norm": 0.6019837260246277, "lr": 3.220299471181898e-06, "epoch": 1, "t": 4.057452917098999}
|
| 279 |
+
{"step": 279, "loss": 1.1457055807113647, "grad_norm": 0.5977408289909363, "lr": 3.035568442875136e-06, "epoch": 1, "t": 4.623497724533081}
|
| 280 |
+
{"step": 280, "loss": 1.2408931255340576, "grad_norm": 0.5990954041481018, "lr": 2.85612772694579e-06, "epoch": 1, "t": 4.593478441238403}
|
| 281 |
+
{"step": 281, "loss": 1.0744915008544922, "grad_norm": 0.675454318523407, "lr": 2.6819975365085237e-06, "epoch": 1, "t": 3.325536012649536}
|
| 282 |
+
{"step": 282, "loss": 1.1200270652770996, "grad_norm": 0.6015303134918213, "lr": 2.5131974864734066e-06, "epoch": 1, "t": 3.777236223220825}
|
| 283 |
+
{"step": 283, "loss": 1.0741581916809082, "grad_norm": 0.5861004590988159, "lr": 2.349746591336405e-06, "epoch": 1, "t": 3.6383676528930664}
|
| 284 |
+
{"step": 284, "loss": 1.1736156940460205, "grad_norm": 0.6229422092437744, "lr": 2.191663263037458e-06, "epoch": 1, "t": 3.4343810081481934}
|
| 285 |
+
{"step": 285, "loss": 0.9877618551254272, "grad_norm": 0.5953364968299866, "lr": 2.0389653088865036e-06, "epoch": 1, "t": 2.8251194953918457}
|
| 286 |
+
{"step": 286, "loss": 1.1036381721496582, "grad_norm": 0.6857370138168335, "lr": 1.8916699295575324e-06, "epoch": 1, "t": 3.7751753330230713}
|
| 287 |
+
{"step": 287, "loss": 1.0825369358062744, "grad_norm": 0.6057208180427551, "lr": 1.7497937171510547e-06, "epoch": 1, "t": 4.018556118011475}
|
| 288 |
+
{"step": 288, "loss": 1.108640193939209, "grad_norm": 0.5292966961860657, "lr": 1.6133526533250565e-06, "epoch": 1, "t": 3.6731085777282715}
|
| 289 |
+
{"step": 289, "loss": 1.2366762161254883, "grad_norm": 0.6128385663032532, "lr": 1.4823621074947503e-06, "epoch": 1, "t": 4.063400745391846}
|
| 290 |
+
{"step": 290, "loss": 1.1423113346099854, "grad_norm": 0.7354625463485718, "lr": 1.3568368351012717e-06, "epoch": 1, "t": 4.498003244400024}
|
| 291 |
+
{"step": 291, "loss": 1.0440839529037476, "grad_norm": 0.5979165434837341, "lr": 1.236790975949592e-06, "epoch": 1, "t": 4.201056480407715}
|
| 292 |
+
{"step": 292, "loss": 1.125098466873169, "grad_norm": 0.6045401692390442, "lr": 1.1222380526156928e-06, "epoch": 1, "t": 3.946053981781006}
|
| 293 |
+
{"step": 293, "loss": 1.001310110092163, "grad_norm": 0.5488253831863403, "lr": 1.0131909689233442e-06, "epoch": 1, "t": 3.8486053943634033}
|
| 294 |
+
{"step": 294, "loss": 1.1466588973999023, "grad_norm": 0.5438467264175415, "lr": 9.096620084905472e-07, "epoch": 1, "t": 4.193962812423706}
|
| 295 |
+
{"step": 295, "loss": 1.0969634056091309, "grad_norm": 0.5312578678131104, "lr": 8.11662833345822e-07, "epoch": 1, "t": 3.706310987472534}
|
| 296 |
+
{"step": 296, "loss": 1.2715599536895752, "grad_norm": 0.7167814373970032, "lr": 7.192044826145771e-07, "epoch": 1, "t": 3.7465150356292725}
|
| 297 |
+
{"step": 297, "loss": 1.0604114532470703, "grad_norm": 0.6269643902778625, "lr": 6.322973712755697e-07, "epoch": 1, "t": 3.352118492126465}
|
| 298 |
+
{"step": 298, "loss": 1.2422224283218384, "grad_norm": 0.602403998374939, "lr": 5.509512889877333e-07, "epoch": 1, "t": 4.141459941864014}
|
| 299 |
+
{"step": 299, "loss": 1.167026162147522, "grad_norm": 0.6394636631011963, "lr": 4.7517539898741524e-07, "epoch": 1, "t": 4.296651601791382}
|
| 300 |
+
{"step": 300, "loss": 1.1355189085006714, "grad_norm": 0.582977294921875, "lr": 4.049782370561583e-07, "epoch": 1, "t": 3.8555452823638916}
|
| 301 |
+
{"step": 301, "loss": 1.163539171218872, "grad_norm": 0.6136392951011658, "lr": 3.4036771055923066e-07, "epoch": 1, "t": 3.868520498275757}
|
| 302 |
+
{"step": 302, "loss": 1.1838977336883545, "grad_norm": 0.6237985491752625, "lr": 2.813510975548772e-07, "epoch": 1, "t": 3.939094066619873}
|
| 303 |
+
{"step": 303, "loss": 1.1358118057250977, "grad_norm": 0.5978771448135376, "lr": 2.2793504597447002e-07, "epoch": 1, "t": 4.025428533554077}
|
| 304 |
+
{"step": 304, "loss": 1.2010633945465088, "grad_norm": 0.6446064710617065, "lr": 1.8012557287367392e-07, "epoch": 1, "t": 3.91121768951416}
|
| 305 |
+
{"step": 305, "loss": 1.1053358316421509, "grad_norm": 0.5842671394348145, "lr": 1.379280637546443e-07, "epoch": 1, "t": 3.6552557945251465}
|
| 306 |
+
{"step": 306, "loss": 1.152795433998108, "grad_norm": 0.5697402358055115, "lr": 1.0134727195937333e-07, "epoch": 1, "t": 3.8585729598999023}
|
| 307 |
+
{"step": 307, "loss": 1.121507167816162, "grad_norm": 0.6407153606414795, "lr": 7.038731813426291e-08, "epoch": 1, "t": 3.782182455062866}
|
| 308 |
+
{"step": 308, "loss": 1.2742674350738525, "grad_norm": 0.5971187353134155, "lr": 4.5051689765929214e-08, "epoch": 1, "t": 4.210731029510498}
|
| 309 |
+
{"step": 309, "loss": 1.0135602951049805, "grad_norm": 0.5833330154418945, "lr": 2.534324078837802e-08, "epoch": 1, "t": 3.4627304077148438}
|
| 310 |
+
{"step": 310, "loss": 1.171689510345459, "grad_norm": 0.5638537406921387, "lr": 1.1264191261528557e-08, "epoch": 1, "t": 4.0617101192474365}
|
| 311 |
+
{"step": 311, "loss": 1.0656862258911133, "grad_norm": 0.6740520000457764, "lr": 2.8161271211024633e-09, "epoch": 1, "t": 3.8180243968963623}
|
| 312 |
+
{"step": 312, "loss": 1.0350477695465088, "grad_norm": 0.5520936846733093, "lr": 0.0, "epoch": 1, "t": 3.6514415740966797}
|
12b/u_top-s0/README.md
ADDED
|
@@ -0,0 +1,206 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: google/gemma-3-12b-it
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- base_model:adapter:google/gemma-3-12b-it
|
| 6 |
+
- lora
|
| 7 |
+
- transformers
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# Model Card for Model ID
|
| 11 |
+
|
| 12 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
## Model Details
|
| 17 |
+
|
| 18 |
+
### Model Description
|
| 19 |
+
|
| 20 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
- **Developed by:** [More Information Needed]
|
| 25 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 26 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 27 |
+
- **Model type:** [More Information Needed]
|
| 28 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 29 |
+
- **License:** [More Information Needed]
|
| 30 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 31 |
+
|
| 32 |
+
### Model Sources [optional]
|
| 33 |
+
|
| 34 |
+
<!-- Provide the basic links for the model. -->
|
| 35 |
+
|
| 36 |
+
- **Repository:** [More Information Needed]
|
| 37 |
+
- **Paper [optional]:** [More Information Needed]
|
| 38 |
+
- **Demo [optional]:** [More Information Needed]
|
| 39 |
+
|
| 40 |
+
## Uses
|
| 41 |
+
|
| 42 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 43 |
+
|
| 44 |
+
### Direct Use
|
| 45 |
+
|
| 46 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 47 |
+
|
| 48 |
+
[More Information Needed]
|
| 49 |
+
|
| 50 |
+
### Downstream Use [optional]
|
| 51 |
+
|
| 52 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 53 |
+
|
| 54 |
+
[More Information Needed]
|
| 55 |
+
|
| 56 |
+
### Out-of-Scope Use
|
| 57 |
+
|
| 58 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 59 |
+
|
| 60 |
+
[More Information Needed]
|
| 61 |
+
|
| 62 |
+
## Bias, Risks, and Limitations
|
| 63 |
+
|
| 64 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 65 |
+
|
| 66 |
+
[More Information Needed]
|
| 67 |
+
|
| 68 |
+
### Recommendations
|
| 69 |
+
|
| 70 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 71 |
+
|
| 72 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 73 |
+
|
| 74 |
+
## How to Get Started with the Model
|
| 75 |
+
|
| 76 |
+
Use the code below to get started with the model.
|
| 77 |
+
|
| 78 |
+
[More Information Needed]
|
| 79 |
+
|
| 80 |
+
## Training Details
|
| 81 |
+
|
| 82 |
+
### Training Data
|
| 83 |
+
|
| 84 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 85 |
+
|
| 86 |
+
[More Information Needed]
|
| 87 |
+
|
| 88 |
+
### Training Procedure
|
| 89 |
+
|
| 90 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 91 |
+
|
| 92 |
+
#### Preprocessing [optional]
|
| 93 |
+
|
| 94 |
+
[More Information Needed]
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
#### Training Hyperparameters
|
| 98 |
+
|
| 99 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 100 |
+
|
| 101 |
+
#### Speeds, Sizes, Times [optional]
|
| 102 |
+
|
| 103 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 104 |
+
|
| 105 |
+
[More Information Needed]
|
| 106 |
+
|
| 107 |
+
## Evaluation
|
| 108 |
+
|
| 109 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 110 |
+
|
| 111 |
+
### Testing Data, Factors & Metrics
|
| 112 |
+
|
| 113 |
+
#### Testing Data
|
| 114 |
+
|
| 115 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 116 |
+
|
| 117 |
+
[More Information Needed]
|
| 118 |
+
|
| 119 |
+
#### Factors
|
| 120 |
+
|
| 121 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 122 |
+
|
| 123 |
+
[More Information Needed]
|
| 124 |
+
|
| 125 |
+
#### Metrics
|
| 126 |
+
|
| 127 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 128 |
+
|
| 129 |
+
[More Information Needed]
|
| 130 |
+
|
| 131 |
+
### Results
|
| 132 |
+
|
| 133 |
+
[More Information Needed]
|
| 134 |
+
|
| 135 |
+
#### Summary
|
| 136 |
+
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
## Model Examination [optional]
|
| 140 |
+
|
| 141 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 142 |
+
|
| 143 |
+
[More Information Needed]
|
| 144 |
+
|
| 145 |
+
## Environmental Impact
|
| 146 |
+
|
| 147 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 148 |
+
|
| 149 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 150 |
+
|
| 151 |
+
- **Hardware Type:** [More Information Needed]
|
| 152 |
+
- **Hours used:** [More Information Needed]
|
| 153 |
+
- **Cloud Provider:** [More Information Needed]
|
| 154 |
+
- **Compute Region:** [More Information Needed]
|
| 155 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 156 |
+
|
| 157 |
+
## Technical Specifications [optional]
|
| 158 |
+
|
| 159 |
+
### Model Architecture and Objective
|
| 160 |
+
|
| 161 |
+
[More Information Needed]
|
| 162 |
+
|
| 163 |
+
### Compute Infrastructure
|
| 164 |
+
|
| 165 |
+
[More Information Needed]
|
| 166 |
+
|
| 167 |
+
#### Hardware
|
| 168 |
+
|
| 169 |
+
[More Information Needed]
|
| 170 |
+
|
| 171 |
+
#### Software
|
| 172 |
+
|
| 173 |
+
[More Information Needed]
|
| 174 |
+
|
| 175 |
+
## Citation [optional]
|
| 176 |
+
|
| 177 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 178 |
+
|
| 179 |
+
**BibTeX:**
|
| 180 |
+
|
| 181 |
+
[More Information Needed]
|
| 182 |
+
|
| 183 |
+
**APA:**
|
| 184 |
+
|
| 185 |
+
[More Information Needed]
|
| 186 |
+
|
| 187 |
+
## Glossary [optional]
|
| 188 |
+
|
| 189 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 190 |
+
|
| 191 |
+
[More Information Needed]
|
| 192 |
+
|
| 193 |
+
## More Information [optional]
|
| 194 |
+
|
| 195 |
+
[More Information Needed]
|
| 196 |
+
|
| 197 |
+
## Model Card Authors [optional]
|
| 198 |
+
|
| 199 |
+
[More Information Needed]
|
| 200 |
+
|
| 201 |
+
## Model Card Contact
|
| 202 |
+
|
| 203 |
+
[More Information Needed]
|
| 204 |
+
### Framework versions
|
| 205 |
+
|
| 206 |
+
- PEFT 0.17.1
|
12b/u_top-s0/adapter_config.json
ADDED
|
@@ -0,0 +1,186 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": {
|
| 4 |
+
"base_model_class": "Gemma3ForConditionalGeneration",
|
| 5 |
+
"parent_library": "transformers.models.gemma3.modeling_gemma3"
|
| 6 |
+
},
|
| 7 |
+
"base_model_name_or_path": "google/gemma-3-12b-it",
|
| 8 |
+
"bias": "none",
|
| 9 |
+
"corda_config": null,
|
| 10 |
+
"eva_config": null,
|
| 11 |
+
"exclude_modules": null,
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": true,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layer_replication": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"layers_to_transform": null,
|
| 18 |
+
"loftq_config": {},
|
| 19 |
+
"lora_alpha": 128,
|
| 20 |
+
"lora_bias": false,
|
| 21 |
+
"lora_dropout": 0.05,
|
| 22 |
+
"megatron_config": null,
|
| 23 |
+
"megatron_core": "megatron.core",
|
| 24 |
+
"modules_to_save": null,
|
| 25 |
+
"peft_type": "LORA",
|
| 26 |
+
"qalora_group_size": 16,
|
| 27 |
+
"r": 64,
|
| 28 |
+
"rank_pattern": {},
|
| 29 |
+
"revision": null,
|
| 30 |
+
"target_modules": [
|
| 31 |
+
"language_model.layers.7.self_attn.k_proj",
|
| 32 |
+
"39.self_attn.k_proj",
|
| 33 |
+
"language_model.layers.13.self_attn.q_proj",
|
| 34 |
+
"47.self_attn.k_proj",
|
| 35 |
+
"language_model.layers.9.self_attn.k_proj",
|
| 36 |
+
"46.self_attn.v_proj",
|
| 37 |
+
"43.self_attn.v_proj",
|
| 38 |
+
"language_model.layers.20.self_attn.v_proj",
|
| 39 |
+
"language_model.layers.5.self_attn.q_proj",
|
| 40 |
+
"42.self_attn.k_proj",
|
| 41 |
+
"language_model.layers.22.self_attn.k_proj",
|
| 42 |
+
"language_model.layers.23.self_attn.k_proj",
|
| 43 |
+
"language_model.layers.17.self_attn.k_proj",
|
| 44 |
+
"language_model.layers.20.self_attn.q_proj",
|
| 45 |
+
"42.self_attn.q_proj",
|
| 46 |
+
"language_model.layers.8.self_attn.k_proj",
|
| 47 |
+
"30.self_attn.v_proj",
|
| 48 |
+
"language_model.layers.0.self_attn.q_proj",
|
| 49 |
+
"31.self_attn.q_proj",
|
| 50 |
+
"30.self_attn.q_proj",
|
| 51 |
+
"language_model.layers.8.self_attn.q_proj",
|
| 52 |
+
"language_model.layers.22.self_attn.q_proj",
|
| 53 |
+
"32.self_attn.k_proj",
|
| 54 |
+
"32.self_attn.v_proj",
|
| 55 |
+
"language_model.layers.25.self_attn.k_proj",
|
| 56 |
+
"39.self_attn.v_proj",
|
| 57 |
+
"language_model.layers.2.self_attn.k_proj",
|
| 58 |
+
"language_model.layers.26.self_attn.q_proj",
|
| 59 |
+
"language_model.layers.25.self_attn.q_proj",
|
| 60 |
+
"35.self_attn.v_proj",
|
| 61 |
+
"gate_proj",
|
| 62 |
+
"language_model.layers.1.self_attn.q_proj",
|
| 63 |
+
"47.self_attn.q_proj",
|
| 64 |
+
"35.self_attn.k_proj",
|
| 65 |
+
"38.self_attn.v_proj",
|
| 66 |
+
"language_model.layers.4.self_attn.k_proj",
|
| 67 |
+
"language_model.layers.24.self_attn.k_proj",
|
| 68 |
+
"language_model.layers.19.self_attn.v_proj",
|
| 69 |
+
"37.self_attn.k_proj",
|
| 70 |
+
"language_model.layers.5.self_attn.v_proj",
|
| 71 |
+
"language_model.layers.16.self_attn.v_proj",
|
| 72 |
+
"language_model.layers.6.self_attn.k_proj",
|
| 73 |
+
"up_proj",
|
| 74 |
+
"29.self_attn.v_proj",
|
| 75 |
+
"27.self_attn.q_proj",
|
| 76 |
+
"43.self_attn.k_proj",
|
| 77 |
+
"language_model.layers.12.self_attn.q_proj",
|
| 78 |
+
"language_model.layers.14.self_attn.q_proj",
|
| 79 |
+
"45.self_attn.v_proj",
|
| 80 |
+
"language_model.layers.23.self_attn.q_proj",
|
| 81 |
+
"42.self_attn.v_proj",
|
| 82 |
+
"34.self_attn.q_proj",
|
| 83 |
+
"41.self_attn.k_proj",
|
| 84 |
+
"language_model.layers.9.self_attn.q_proj",
|
| 85 |
+
"language_model.layers.11.self_attn.v_proj",
|
| 86 |
+
"down_proj",
|
| 87 |
+
"language_model.layers.14.self_attn.v_proj",
|
| 88 |
+
"language_model.layers.22.self_attn.v_proj",
|
| 89 |
+
"language_model.layers.24.self_attn.q_proj",
|
| 90 |
+
"language_model.layers.11.self_attn.k_proj",
|
| 91 |
+
"language_model.layers.2.self_attn.v_proj",
|
| 92 |
+
"language_model.layers.15.self_attn.q_proj",
|
| 93 |
+
"language_model.layers.13.self_attn.v_proj",
|
| 94 |
+
"30.self_attn.k_proj",
|
| 95 |
+
"41.self_attn.q_proj",
|
| 96 |
+
"language_model.layers.7.self_attn.v_proj",
|
| 97 |
+
"46.self_attn.q_proj",
|
| 98 |
+
"36.self_attn.k_proj",
|
| 99 |
+
"47.self_attn.v_proj",
|
| 100 |
+
"language_model.layers.15.self_attn.v_proj",
|
| 101 |
+
"language_model.layers.11.self_attn.q_proj",
|
| 102 |
+
"41.self_attn.v_proj",
|
| 103 |
+
"43.self_attn.q_proj",
|
| 104 |
+
"language_model.layers.26.self_attn.v_proj",
|
| 105 |
+
"language_model.layers.13.self_attn.k_proj",
|
| 106 |
+
"language_model.layers.24.self_attn.v_proj",
|
| 107 |
+
"37.self_attn.v_proj",
|
| 108 |
+
"45.self_attn.k_proj",
|
| 109 |
+
"language_model.layers.18.self_attn.q_proj",
|
| 110 |
+
"36.self_attn.v_proj",
|
| 111 |
+
"o_proj",
|
| 112 |
+
"29.self_attn.q_proj",
|
| 113 |
+
"language_model.layers.21.self_attn.k_proj",
|
| 114 |
+
"language_model.layers.5.self_attn.k_proj",
|
| 115 |
+
"language_model.layers.26.self_attn.k_proj",
|
| 116 |
+
"language_model.layers.14.self_attn.k_proj",
|
| 117 |
+
"language_model.layers.8.self_attn.v_proj",
|
| 118 |
+
"language_model.layers.0.self_attn.k_proj",
|
| 119 |
+
"40.self_attn.q_proj",
|
| 120 |
+
"language_model.layers.3.self_attn.v_proj",
|
| 121 |
+
"44.self_attn.k_proj",
|
| 122 |
+
"language_model.layers.10.self_attn.k_proj",
|
| 123 |
+
"language_model.layers.16.self_attn.q_proj",
|
| 124 |
+
"39.self_attn.q_proj",
|
| 125 |
+
"33.self_attn.q_proj",
|
| 126 |
+
"language_model.layers.18.self_attn.v_proj",
|
| 127 |
+
"language_model.layers.16.self_attn.k_proj",
|
| 128 |
+
"45.self_attn.q_proj",
|
| 129 |
+
"34.self_attn.v_proj",
|
| 130 |
+
"language_model.layers.25.self_attn.v_proj",
|
| 131 |
+
"language_model.layers.19.self_attn.k_proj",
|
| 132 |
+
"35.self_attn.q_proj",
|
| 133 |
+
"31.self_attn.v_proj",
|
| 134 |
+
"language_model.layers.0.self_attn.v_proj",
|
| 135 |
+
"language_model.layers.19.self_attn.q_proj",
|
| 136 |
+
"28.self_attn.v_proj",
|
| 137 |
+
"language_model.layers.10.self_attn.v_proj",
|
| 138 |
+
"36.self_attn.q_proj",
|
| 139 |
+
"language_model.layers.18.self_attn.k_proj",
|
| 140 |
+
"40.self_attn.v_proj",
|
| 141 |
+
"32.self_attn.q_proj",
|
| 142 |
+
"language_model.layers.6.self_attn.q_proj",
|
| 143 |
+
"language_model.layers.2.self_attn.q_proj",
|
| 144 |
+
"language_model.layers.7.self_attn.q_proj",
|
| 145 |
+
"language_model.layers.6.self_attn.v_proj",
|
| 146 |
+
"28.self_attn.q_proj",
|
| 147 |
+
"language_model.layers.4.self_attn.q_proj",
|
| 148 |
+
"38.self_attn.k_proj",
|
| 149 |
+
"27.self_attn.k_proj",
|
| 150 |
+
"language_model.layers.3.self_attn.k_proj",
|
| 151 |
+
"37.self_attn.q_proj",
|
| 152 |
+
"46.self_attn.k_proj",
|
| 153 |
+
"language_model.layers.1.self_attn.k_proj",
|
| 154 |
+
"language_model.layers.10.self_attn.q_proj",
|
| 155 |
+
"33.self_attn.k_proj",
|
| 156 |
+
"language_model.layers.17.self_attn.v_proj",
|
| 157 |
+
"language_model.layers.21.self_attn.q_proj",
|
| 158 |
+
"31.self_attn.k_proj",
|
| 159 |
+
"language_model.layers.17.self_attn.q_proj",
|
| 160 |
+
"language_model.layers.3.self_attn.q_proj",
|
| 161 |
+
"38.self_attn.q_proj",
|
| 162 |
+
"language_model.layers.1.self_attn.v_proj",
|
| 163 |
+
"language_model.layers.15.self_attn.k_proj",
|
| 164 |
+
"language_model.layers.23.self_attn.v_proj",
|
| 165 |
+
"language_model.layers.4.self_attn.v_proj",
|
| 166 |
+
"language_model.layers.12.self_attn.k_proj",
|
| 167 |
+
"34.self_attn.k_proj",
|
| 168 |
+
"27.self_attn.v_proj",
|
| 169 |
+
"44.self_attn.q_proj",
|
| 170 |
+
"28.self_attn.k_proj",
|
| 171 |
+
"33.self_attn.v_proj",
|
| 172 |
+
"44.self_attn.v_proj",
|
| 173 |
+
"29.self_attn.k_proj",
|
| 174 |
+
"40.self_attn.k_proj",
|
| 175 |
+
"language_model.layers.9.self_attn.v_proj",
|
| 176 |
+
"language_model.layers.21.self_attn.v_proj",
|
| 177 |
+
"language_model.layers.12.self_attn.v_proj",
|
| 178 |
+
"language_model.layers.20.self_attn.k_proj"
|
| 179 |
+
],
|
| 180 |
+
"target_parameters": null,
|
| 181 |
+
"task_type": null,
|
| 182 |
+
"trainable_token_indices": null,
|
| 183 |
+
"use_dora": false,
|
| 184 |
+
"use_qalora": false,
|
| 185 |
+
"use_rslora": false
|
| 186 |
+
}
|
12b/u_top-s0/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:df7c7b8cab8246a97653b9c5a771aef08e370cf04d0c91a45cc2ca6291bfba3c
|
| 3 |
+
size 1047628488
|
12b/u_top-s0/config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "google/gemma-3-12b-it",
|
| 3 |
+
"data": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma12/arms/TASKW-u_top-gemma12.jsonl",
|
| 4 |
+
"run": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma12/runs/u_top-s0",
|
| 5 |
+
"steps": 0,
|
| 6 |
+
"batch": 16,
|
| 7 |
+
"micro": 1,
|
| 8 |
+
"seq_len": 4096,
|
| 9 |
+
"lr": 0.0001,
|
| 10 |
+
"warmup": 10,
|
| 11 |
+
"lora_rank": 64,
|
| 12 |
+
"lora_targets": "all-linear",
|
| 13 |
+
"lora_targets_file": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/code/gemma3_12b_lora_targets.json",
|
| 14 |
+
"seed": 0,
|
| 15 |
+
"bos": true,
|
| 16 |
+
"no_merge": false,
|
| 17 |
+
"save_at": "",
|
| 18 |
+
"replay": "",
|
| 19 |
+
"replay_frac": 0.0,
|
| 20 |
+
"constant_lr": false,
|
| 21 |
+
"schedule": "hf-cosine",
|
| 22 |
+
"min_lr_frac": 0.0,
|
| 23 |
+
"warmup_ratio": 0.05,
|
| 24 |
+
"lora_dropout": 0.05,
|
| 25 |
+
"no_pack": true,
|
| 26 |
+
"chat_sft": true,
|
| 27 |
+
"data_frac": 1.0,
|
| 28 |
+
"base_adapter": "",
|
| 29 |
+
"ckpt_every": 0,
|
| 30 |
+
"kill_after": 0,
|
| 31 |
+
"method": "doc_lm_lora",
|
| 32 |
+
"world": 2
|
| 33 |
+
}
|
12b/u_top-s0/log.jsonl
ADDED
|
@@ -0,0 +1,312 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 1, "loss": 2.327320098876953, "grad_norm": 4.479649543762207, "lr": 1.25e-05, "epoch": 1, "t": 3.2861361503601074}
|
| 2 |
+
{"step": 2, "loss": 2.2081398963928223, "grad_norm": 4.171542644500732, "lr": 1.8750000000000002e-05, "epoch": 1, "t": 3.501610279083252}
|
| 3 |
+
{"step": 3, "loss": 2.227512836456299, "grad_norm": 3.7014870643615723, "lr": 2.5e-05, "epoch": 1, "t": 3.3065903186798096}
|
| 4 |
+
{"step": 4, "loss": 2.0284104347229004, "grad_norm": 3.0142970085144043, "lr": 3.125e-05, "epoch": 1, "t": 2.5298056602478027}
|
| 5 |
+
{"step": 5, "loss": 1.8827426433563232, "grad_norm": 2.323382616043091, "lr": 3.7500000000000003e-05, "epoch": 1, "t": 3.4053468704223633}
|
| 6 |
+
{"step": 6, "loss": 1.7148656845092773, "grad_norm": 1.4067871570587158, "lr": 4.375e-05, "epoch": 1, "t": 3.2648541927337646}
|
| 7 |
+
{"step": 7, "loss": 1.6916370391845703, "grad_norm": 1.0447163581848145, "lr": 5e-05, "epoch": 1, "t": 2.7632486820220947}
|
| 8 |
+
{"step": 8, "loss": 1.4887657165527344, "grad_norm": 0.8689322471618652, "lr": 5.6250000000000005e-05, "epoch": 1, "t": 2.8857762813568115}
|
| 9 |
+
{"step": 9, "loss": 1.5974476337432861, "grad_norm": 0.9982807040214539, "lr": 6.25e-05, "epoch": 1, "t": 3.258718729019165}
|
| 10 |
+
{"step": 10, "loss": 1.6818633079528809, "grad_norm": 1.0472453832626343, "lr": 6.875e-05, "epoch": 1, "t": 3.62207293510437}
|
| 11 |
+
{"step": 11, "loss": 1.4479854106903076, "grad_norm": 0.9008411169052124, "lr": 7.500000000000001e-05, "epoch": 1, "t": 2.7511041164398193}
|
| 12 |
+
{"step": 12, "loss": 1.4024055004119873, "grad_norm": 0.8531894087791443, "lr": 8.125000000000001e-05, "epoch": 1, "t": 2.960697889328003}
|
| 13 |
+
{"step": 13, "loss": 1.2542465925216675, "grad_norm": 0.7195949554443359, "lr": 8.75e-05, "epoch": 1, "t": 2.882171869277954}
|
| 14 |
+
{"step": 14, "loss": 1.5960586071014404, "grad_norm": 0.7588453888893127, "lr": 9.375e-05, "epoch": 1, "t": 2.744157075881958}
|
| 15 |
+
{"step": 15, "loss": 1.5088458061218262, "grad_norm": 0.7630959749221802, "lr": 0.0001, "epoch": 1, "t": 2.864926338195801}
|
| 16 |
+
{"step": 16, "loss": 1.4294493198394775, "grad_norm": 0.75345778465271, "lr": 0.0001, "epoch": 1, "t": 3.0488786697387695}
|
| 17 |
+
{"step": 17, "loss": 1.3520256280899048, "grad_norm": 0.7315390706062317, "lr": 9.99971838728789e-05, "epoch": 1, "t": 3.106358528137207}
|
| 18 |
+
{"step": 18, "loss": 1.3143670558929443, "grad_norm": 0.8493432998657227, "lr": 9.998873580873848e-05, "epoch": 1, "t": 2.571648597717285}
|
| 19 |
+
{"step": 19, "loss": 1.3237378597259521, "grad_norm": 0.7623825669288635, "lr": 9.997465675921163e-05, "epoch": 1, "t": 3.074885845184326}
|
| 20 |
+
{"step": 20, "loss": 1.321047067642212, "grad_norm": 0.6148121356964111, "lr": 9.995494831023409e-05, "epoch": 1, "t": 2.9477245807647705}
|
| 21 |
+
{"step": 21, "loss": 1.2125921249389648, "grad_norm": 0.6126531958580017, "lr": 9.992961268186573e-05, "epoch": 1, "t": 3.16160249710083}
|
| 22 |
+
{"step": 22, "loss": 1.3310753107070923, "grad_norm": 0.6271969079971313, "lr": 9.989865272804063e-05, "epoch": 1, "t": 3.856884479522705}
|
| 23 |
+
{"step": 23, "loss": 1.0960006713867188, "grad_norm": 0.5843579173088074, "lr": 9.986207193624536e-05, "epoch": 1, "t": 3.6228508949279785}
|
| 24 |
+
{"step": 24, "loss": 1.3347666263580322, "grad_norm": 0.5616052746772766, "lr": 9.981987442712633e-05, "epoch": 1, "t": 2.879272699356079}
|
| 25 |
+
{"step": 25, "loss": 1.156449794769287, "grad_norm": 0.5829220414161682, "lr": 9.977206495402554e-05, "epoch": 1, "t": 2.981518030166626}
|
| 26 |
+
{"step": 26, "loss": 1.3228797912597656, "grad_norm": 0.6559152603149414, "lr": 9.971864890244513e-05, "epoch": 1, "t": 3.409393787384033}
|
| 27 |
+
{"step": 27, "loss": 1.4145439863204956, "grad_norm": 0.6042874455451965, "lr": 9.965963228944078e-05, "epoch": 1, "t": 3.3931119441986084}
|
| 28 |
+
{"step": 28, "loss": 1.2591763734817505, "grad_norm": 0.6486919522285461, "lr": 9.959502176294383e-05, "epoch": 1, "t": 3.559197187423706}
|
| 29 |
+
{"step": 29, "loss": 1.2655839920043945, "grad_norm": 0.7721403241157532, "lr": 9.95248246010126e-05, "epoch": 1, "t": 3.674147129058838}
|
| 30 |
+
{"step": 30, "loss": 1.3437391519546509, "grad_norm": 0.5789604187011719, "lr": 9.944904871101228e-05, "epoch": 1, "t": 3.036595344543457}
|
| 31 |
+
{"step": 31, "loss": 1.313507318496704, "grad_norm": 0.6066290736198425, "lr": 9.936770262872443e-05, "epoch": 1, "t": 2.792367935180664}
|
| 32 |
+
{"step": 32, "loss": 1.2701749801635742, "grad_norm": 0.5581668615341187, "lr": 9.928079551738543e-05, "epoch": 1, "t": 3.64007568359375}
|
| 33 |
+
{"step": 33, "loss": 1.2571465969085693, "grad_norm": 0.6207336783409119, "lr": 9.918833716665419e-05, "epoch": 1, "t": 2.840121030807495}
|
| 34 |
+
{"step": 34, "loss": 1.1672500371932983, "grad_norm": 0.6164065599441528, "lr": 9.909033799150946e-05, "epoch": 1, "t": 3.0070223808288574}
|
| 35 |
+
{"step": 35, "loss": 1.1275960206985474, "grad_norm": 0.6311043500900269, "lr": 9.898680903107666e-05, "epoch": 1, "t": 2.7110443115234375}
|
| 36 |
+
{"step": 36, "loss": 1.2309033870697021, "grad_norm": 0.5920257568359375, "lr": 9.887776194738432e-05, "epoch": 1, "t": 3.2392303943634033}
|
| 37 |
+
{"step": 37, "loss": 1.26517653465271, "grad_norm": 0.609675943851471, "lr": 9.876320902405042e-05, "epoch": 1, "t": 3.149461269378662}
|
| 38 |
+
{"step": 38, "loss": 1.2478092908859253, "grad_norm": 0.6141680479049683, "lr": 9.864316316489873e-05, "epoch": 1, "t": 2.834200143814087}
|
| 39 |
+
{"step": 39, "loss": 1.3558597564697266, "grad_norm": 0.6106985211372375, "lr": 9.851763789250525e-05, "epoch": 1, "t": 2.8861913681030273}
|
| 40 |
+
{"step": 40, "loss": 1.225052833557129, "grad_norm": 0.590245246887207, "lr": 9.838664734667495e-05, "epoch": 1, "t": 2.9458816051483154}
|
| 41 |
+
{"step": 41, "loss": 1.1565661430358887, "grad_norm": 0.5631913542747498, "lr": 9.825020628284896e-05, "epoch": 1, "t": 2.8674819469451904}
|
| 42 |
+
{"step": 42, "loss": 1.2926607131958008, "grad_norm": 0.6024816632270813, "lr": 9.810833007044247e-05, "epoch": 1, "t": 2.8970460891723633}
|
| 43 |
+
{"step": 43, "loss": 1.345505952835083, "grad_norm": 0.5541171431541443, "lr": 9.796103469111351e-05, "epoch": 1, "t": 3.6500465869903564}
|
| 44 |
+
{"step": 44, "loss": 1.2472412586212158, "grad_norm": 0.6114717721939087, "lr": 9.780833673696254e-05, "epoch": 1, "t": 3.0729925632476807}
|
| 45 |
+
{"step": 45, "loss": 1.1458070278167725, "grad_norm": 0.5363909602165222, "lr": 9.76502534086636e-05, "epoch": 1, "t": 2.9130849838256836}
|
| 46 |
+
{"step": 46, "loss": 1.2604302167892456, "grad_norm": 0.5959209203720093, "lr": 9.74868025135266e-05, "epoch": 1, "t": 3.0661916732788086}
|
| 47 |
+
{"step": 47, "loss": 1.293121576309204, "grad_norm": 0.618234395980835, "lr": 9.731800246349148e-05, "epoch": 1, "t": 3.0375189781188965}
|
| 48 |
+
{"step": 48, "loss": 1.212491750717163, "grad_norm": 0.5881310105323792, "lr": 9.714387227305422e-05, "epoch": 1, "t": 2.948955535888672}
|
| 49 |
+
{"step": 49, "loss": 1.2408185005187988, "grad_norm": 0.5597432255744934, "lr": 9.696443155712486e-05, "epoch": 1, "t": 3.847395181655884}
|
| 50 |
+
{"step": 50, "loss": 1.266973853111267, "grad_norm": 0.609494149684906, "lr": 9.67797005288181e-05, "epoch": 1, "t": 3.2872307300567627}
|
| 51 |
+
{"step": 51, "loss": 1.3280231952667236, "grad_norm": 0.5787083506584167, "lr": 9.65896999971763e-05, "epoch": 1, "t": 3.095487356185913}
|
| 52 |
+
{"step": 52, "loss": 1.235647201538086, "grad_norm": 0.6278812289237976, "lr": 9.639445136482548e-05, "epoch": 1, "t": 2.957137107849121}
|
| 53 |
+
{"step": 53, "loss": 1.2202039957046509, "grad_norm": 0.6624301671981812, "lr": 9.619397662556435e-05, "epoch": 1, "t": 2.855740547180176}
|
| 54 |
+
{"step": 54, "loss": 1.2952053546905518, "grad_norm": 0.6566532254219055, "lr": 9.598829836188694e-05, "epoch": 1, "t": 2.848402738571167}
|
| 55 |
+
{"step": 55, "loss": 1.2492526769638062, "grad_norm": 0.6289007663726807, "lr": 9.577743974243874e-05, "epoch": 1, "t": 2.9781887531280518}
|
| 56 |
+
{"step": 56, "loss": 1.2058625221252441, "grad_norm": 0.594464898109436, "lr": 9.55614245194068e-05, "epoch": 1, "t": 3.2028605937957764}
|
| 57 |
+
{"step": 57, "loss": 1.2552945613861084, "grad_norm": 0.6747407913208008, "lr": 9.534027702584425e-05, "epoch": 1, "t": 3.255425453186035}
|
| 58 |
+
{"step": 58, "loss": 1.1330509185791016, "grad_norm": 0.5906953811645508, "lr": 9.511402217292926e-05, "epoch": 1, "t": 2.984585762023926}
|
| 59 |
+
{"step": 59, "loss": 1.2708070278167725, "grad_norm": 0.6074764728546143, "lr": 9.488268544715896e-05, "epoch": 1, "t": 3.0130326747894287}
|
| 60 |
+
{"step": 60, "loss": 1.1431752443313599, "grad_norm": 0.7789663076400757, "lr": 9.464629290747842e-05, "epoch": 1, "t": 3.494136333465576}
|
| 61 |
+
{"step": 61, "loss": 1.4228079319000244, "grad_norm": 0.6221870183944702, "lr": 9.440487118234535e-05, "epoch": 1, "t": 3.056659460067749}
|
| 62 |
+
{"step": 62, "loss": 1.2959685325622559, "grad_norm": 0.5757694244384766, "lr": 9.415844746673047e-05, "epoch": 1, "t": 2.6466715335845947}
|
| 63 |
+
{"step": 63, "loss": 1.2407996654510498, "grad_norm": 0.5622316598892212, "lr": 9.390704951905411e-05, "epoch": 1, "t": 3.0520288944244385}
|
| 64 |
+
{"step": 64, "loss": 1.3233704566955566, "grad_norm": 0.6001541018486023, "lr": 9.365070565805941e-05, "epoch": 1, "t": 3.242171049118042}
|
| 65 |
+
{"step": 65, "loss": 1.1650807857513428, "grad_norm": 0.5802739262580872, "lr": 9.338944475962237e-05, "epoch": 1, "t": 3.045684337615967}
|
| 66 |
+
{"step": 66, "loss": 1.2278456687927246, "grad_norm": 0.7396714091300964, "lr": 9.312329625349902e-05, "epoch": 1, "t": 2.904010057449341}
|
| 67 |
+
{"step": 67, "loss": 1.2978794574737549, "grad_norm": 0.5690197348594666, "lr": 9.285229012001047e-05, "epoch": 1, "t": 3.4255576133728027}
|
| 68 |
+
{"step": 68, "loss": 1.1597262620925903, "grad_norm": 0.5632451772689819, "lr": 9.257645688666556e-05, "epoch": 1, "t": 2.901581287384033}
|
| 69 |
+
{"step": 69, "loss": 1.2125627994537354, "grad_norm": 0.6128668189048767, "lr": 9.22958276247223e-05, "epoch": 1, "t": 2.7937471866607666}
|
| 70 |
+
{"step": 70, "loss": 1.343188762664795, "grad_norm": 0.5677367448806763, "lr": 9.201043394568773e-05, "epoch": 1, "t": 3.5425591468811035}
|
| 71 |
+
{"step": 71, "loss": 1.0806968212127686, "grad_norm": 0.6341626048088074, "lr": 9.172030799775699e-05, "epoch": 1, "t": 2.7552688121795654}
|
| 72 |
+
{"step": 72, "loss": 1.2531797885894775, "grad_norm": 0.5500874519348145, "lr": 9.142548246219212e-05, "epoch": 1, "t": 3.063446521759033}
|
| 73 |
+
{"step": 73, "loss": 1.1263126134872437, "grad_norm": 0.6508458852767944, "lr": 9.112599054964057e-05, "epoch": 1, "t": 2.8892641067504883}
|
| 74 |
+
{"step": 74, "loss": 1.1582095623016357, "grad_norm": 0.6102361083030701, "lr": 9.082186599639428e-05, "epoch": 1, "t": 2.879180908203125}
|
| 75 |
+
{"step": 75, "loss": 1.179563045501709, "grad_norm": 0.5993869304656982, "lr": 9.051314306058933e-05, "epoch": 1, "t": 3.0629258155822754}
|
| 76 |
+
{"step": 76, "loss": 1.18906569480896, "grad_norm": 0.6275805234909058, "lr": 9.019985651834703e-05, "epoch": 1, "t": 3.01584529876709}
|
| 77 |
+
{"step": 77, "loss": 1.1941587924957275, "grad_norm": 0.5857914686203003, "lr": 8.988204165985649e-05, "epoch": 1, "t": 3.4651057720184326}
|
| 78 |
+
{"step": 78, "loss": 1.2205612659454346, "grad_norm": 0.6189649701118469, "lr": 8.955973428539944e-05, "epoch": 1, "t": 2.9542999267578125}
|
| 79 |
+
{"step": 79, "loss": 1.1894892454147339, "grad_norm": 0.5463715195655823, "lr": 8.923297070131737e-05, "epoch": 1, "t": 3.250633478164673}
|
| 80 |
+
{"step": 80, "loss": 1.1691644191741943, "grad_norm": 0.6536669731140137, "lr": 8.890178771592199e-05, "epoch": 1, "t": 3.5509772300720215}
|
| 81 |
+
{"step": 81, "loss": 1.1745002269744873, "grad_norm": 0.6115691661834717, "lr": 8.856622263534875e-05, "epoch": 1, "t": 3.4411978721618652}
|
| 82 |
+
{"step": 82, "loss": 1.271688461303711, "grad_norm": 0.6372178792953491, "lr": 8.822631325935463e-05, "epoch": 1, "t": 3.5441648960113525}
|
| 83 |
+
{"step": 83, "loss": 1.3316903114318848, "grad_norm": 0.5968702435493469, "lr": 8.788209787706015e-05, "epoch": 1, "t": 4.230627536773682}
|
| 84 |
+
{"step": 84, "loss": 1.1684809923171997, "grad_norm": 0.637873113155365, "lr": 8.753361526263621e-05, "epoch": 1, "t": 3.8836686611175537}
|
| 85 |
+
{"step": 85, "loss": 1.02353036403656, "grad_norm": 0.6712408661842346, "lr": 8.718090467093654e-05, "epoch": 1, "t": 3.0928750038146973}
|
| 86 |
+
{"step": 86, "loss": 1.0785856246948242, "grad_norm": 0.5503003597259521, "lr": 8.682400583307562e-05, "epoch": 1, "t": 2.9400875568389893}
|
| 87 |
+
{"step": 87, "loss": 1.3736032247543335, "grad_norm": 0.5776317119598389, "lr": 8.646295895195333e-05, "epoch": 1, "t": 3.0470528602600098}
|
| 88 |
+
{"step": 88, "loss": 1.028544545173645, "grad_norm": 0.6153590083122253, "lr": 8.609780469772623e-05, "epoch": 1, "t": 2.8241100311279297}
|
| 89 |
+
{"step": 89, "loss": 1.234584927558899, "grad_norm": 0.5607485771179199, "lr": 8.572858420322627e-05, "epoch": 1, "t": 2.953537702560425}
|
| 90 |
+
{"step": 90, "loss": 1.2346268892288208, "grad_norm": 0.5581977367401123, "lr": 8.535533905932738e-05, "epoch": 1, "t": 3.8440258502960205}
|
| 91 |
+
{"step": 91, "loss": 1.221160650253296, "grad_norm": 0.5699642896652222, "lr": 8.497811131026046e-05, "epoch": 1, "t": 4.0340259075164795}
|
| 92 |
+
{"step": 92, "loss": 1.2078444957733154, "grad_norm": 0.5951607823371887, "lr": 8.459694344887732e-05, "epoch": 1, "t": 3.017730712890625}
|
| 93 |
+
{"step": 93, "loss": 1.1464855670928955, "grad_norm": 0.6039294600486755, "lr": 8.421187841186402e-05, "epoch": 1, "t": 2.7073140144348145}
|
| 94 |
+
{"step": 94, "loss": 1.305588960647583, "grad_norm": 0.531472384929657, "lr": 8.382295957490436e-05, "epoch": 1, "t": 3.342186212539673}
|
| 95 |
+
{"step": 95, "loss": 1.2411593198776245, "grad_norm": 0.5925220251083374, "lr": 8.343023074779368e-05, "epoch": 1, "t": 3.294947862625122}
|
| 96 |
+
{"step": 96, "loss": 1.2272279262542725, "grad_norm": 0.6295038461685181, "lr": 8.303373616950408e-05, "epoch": 1, "t": 3.0799553394317627}
|
| 97 |
+
{"step": 97, "loss": 1.2100359201431274, "grad_norm": 0.9218403697013855, "lr": 8.263352050320094e-05, "epoch": 1, "t": 3.069793224334717}
|
| 98 |
+
{"step": 98, "loss": 1.1928156614303589, "grad_norm": 0.5811665058135986, "lr": 8.222962883121196e-05, "epoch": 1, "t": 3.0220353603363037}
|
| 99 |
+
{"step": 99, "loss": 1.2326633930206299, "grad_norm": 0.6870810389518738, "lr": 8.182210664994878e-05, "epoch": 1, "t": 2.5364246368408203}
|
| 100 |
+
{"step": 100, "loss": 1.1224595308303833, "grad_norm": 0.5607088208198547, "lr": 8.141099986478212e-05, "epoch": 1, "t": 3.1791491508483887}
|
| 101 |
+
{"step": 101, "loss": 1.1015534400939941, "grad_norm": 0.5766206979751587, "lr": 8.099635478487064e-05, "epoch": 1, "t": 3.582705020904541}
|
| 102 |
+
{"step": 102, "loss": 1.1344712972640991, "grad_norm": 0.5696788430213928, "lr": 8.057821811794458e-05, "epoch": 1, "t": 3.8053183555603027}
|
| 103 |
+
{"step": 103, "loss": 1.1657072305679321, "grad_norm": 0.6360107064247131, "lr": 8.015663696504422e-05, "epoch": 1, "t": 3.0161685943603516}
|
| 104 |
+
{"step": 104, "loss": 1.0946083068847656, "grad_norm": 0.5309562087059021, "lr": 7.973165881521434e-05, "epoch": 1, "t": 4.169360876083374}
|
| 105 |
+
{"step": 105, "loss": 1.1223301887512207, "grad_norm": 0.6696287989616394, "lr": 7.930333154015466e-05, "epoch": 1, "t": 3.631697416305542}
|
| 106 |
+
{"step": 106, "loss": 1.2191174030303955, "grad_norm": 0.5754199624061584, "lr": 7.88717033888274e-05, "epoch": 1, "t": 2.7675325870513916}
|
| 107 |
+
{"step": 107, "loss": 1.211332082748413, "grad_norm": 0.646818220615387, "lr": 7.843682298202235e-05, "epoch": 1, "t": 2.8578171730041504}
|
| 108 |
+
{"step": 108, "loss": 1.2114726305007935, "grad_norm": 0.6245731711387634, "lr": 7.799873930687978e-05, "epoch": 1, "t": 2.990239143371582}
|
| 109 |
+
{"step": 109, "loss": 1.279434323310852, "grad_norm": 0.5976760387420654, "lr": 7.755750171137246e-05, "epoch": 1, "t": 3.264981746673584}
|
| 110 |
+
{"step": 110, "loss": 1.245403528213501, "grad_norm": 0.6508705019950867, "lr": 7.711315989874677e-05, "epoch": 1, "t": 2.827348470687866}
|
| 111 |
+
{"step": 111, "loss": 1.24924635887146, "grad_norm": 0.5850361585617065, "lr": 7.666576392192389e-05, "epoch": 1, "t": 3.0938448905944824}
|
| 112 |
+
{"step": 112, "loss": 1.0820833444595337, "grad_norm": 0.6018866896629333, "lr": 7.621536417786159e-05, "epoch": 1, "t": 2.8252952098846436}
|
| 113 |
+
{"step": 113, "loss": 1.273302674293518, "grad_norm": 0.6372866630554199, "lr": 7.576201140187727e-05, "epoch": 1, "t": 2.571823835372925}
|
| 114 |
+
{"step": 114, "loss": 1.1704981327056885, "grad_norm": 0.5573928356170654, "lr": 7.530575666193283e-05, "epoch": 1, "t": 3.1482512950897217}
|
| 115 |
+
{"step": 115, "loss": 1.1771819591522217, "grad_norm": 0.6436694264411926, "lr": 7.484665135288213e-05, "epoch": 1, "t": 2.8885557651519775}
|
| 116 |
+
{"step": 116, "loss": 1.045685052871704, "grad_norm": 0.6081012487411499, "lr": 7.438474719068173e-05, "epoch": 1, "t": 2.722137212753296}
|
| 117 |
+
{"step": 117, "loss": 1.1190414428710938, "grad_norm": 0.7006279826164246, "lr": 7.392009620656513e-05, "epoch": 1, "t": 3.499424934387207}
|
| 118 |
+
{"step": 118, "loss": 1.1154733896255493, "grad_norm": 0.7012538313865662, "lr": 7.345275074118185e-05, "epoch": 1, "t": 3.7546942234039307}
|
| 119 |
+
{"step": 119, "loss": 1.1233445405960083, "grad_norm": 0.5250251889228821, "lr": 7.298276343870151e-05, "epoch": 1, "t": 3.8036599159240723}
|
| 120 |
+
{"step": 120, "loss": 1.1060562133789062, "grad_norm": 0.5347779393196106, "lr": 7.251018724088367e-05, "epoch": 1, "t": 2.9217615127563477}
|
| 121 |
+
{"step": 121, "loss": 1.063027262687683, "grad_norm": 0.6037988066673279, "lr": 7.203507538111423e-05, "epoch": 1, "t": 2.8044254779815674}
|
| 122 |
+
{"step": 122, "loss": 1.1519885063171387, "grad_norm": 0.6173638701438904, "lr": 7.155748137840892e-05, "epoch": 1, "t": 3.165515422821045}
|
| 123 |
+
{"step": 123, "loss": 1.1948976516723633, "grad_norm": 0.569068193435669, "lr": 7.107745903138472e-05, "epoch": 1, "t": 2.971468210220337}
|
| 124 |
+
{"step": 124, "loss": 1.233440637588501, "grad_norm": 0.5795004963874817, "lr": 7.059506241219965e-05, "epoch": 1, "t": 3.1745150089263916}
|
| 125 |
+
{"step": 125, "loss": 1.0947480201721191, "grad_norm": 0.595154881477356, "lr": 7.011034586046176e-05, "epoch": 1, "t": 2.9281632900238037}
|
| 126 |
+
{"step": 126, "loss": 1.1543587446212769, "grad_norm": 0.6121411919593811, "lr": 6.962336397710819e-05, "epoch": 1, "t": 2.9809746742248535}
|
| 127 |
+
{"step": 127, "loss": 1.1404320001602173, "grad_norm": 0.6786699891090393, "lr": 6.91341716182545e-05, "epoch": 1, "t": 2.6360747814178467}
|
| 128 |
+
{"step": 128, "loss": 1.155851125717163, "grad_norm": 0.6679185628890991, "lr": 6.864282388901544e-05, "epoch": 1, "t": 2.8419406414031982}
|
| 129 |
+
{"step": 129, "loss": 1.1627235412597656, "grad_norm": 0.5989121198654175, "lr": 6.814937613729766e-05, "epoch": 1, "t": 3.0893776416778564}
|
| 130 |
+
{"step": 130, "loss": 1.1237167119979858, "grad_norm": 0.6347812414169312, "lr": 6.765388394756504e-05, "epoch": 1, "t": 2.6416707038879395}
|
| 131 |
+
{"step": 131, "loss": 1.1113510131835938, "grad_norm": 0.5990824103355408, "lr": 6.715640313457733e-05, "epoch": 1, "t": 3.090550422668457}
|
| 132 |
+
{"step": 132, "loss": 1.1517996788024902, "grad_norm": 0.5490580201148987, "lr": 6.665698973710288e-05, "epoch": 1, "t": 3.0941805839538574}
|
| 133 |
+
{"step": 133, "loss": 1.0341966152191162, "grad_norm": 0.5785680413246155, "lr": 6.615570001160626e-05, "epoch": 1, "t": 2.6230223178863525}
|
| 134 |
+
{"step": 134, "loss": 1.1920065879821777, "grad_norm": 1.1524473428726196, "lr": 6.565259042591113e-05, "epoch": 1, "t": 3.7972874641418457}
|
| 135 |
+
{"step": 135, "loss": 1.1461162567138672, "grad_norm": 0.6095324158668518, "lr": 6.514771765283942e-05, "epoch": 1, "t": 2.835585355758667}
|
| 136 |
+
{"step": 136, "loss": 1.1170728206634521, "grad_norm": 0.6389683485031128, "lr": 6.464113856382752e-05, "epoch": 1, "t": 2.8336169719696045}
|
| 137 |
+
{"step": 137, "loss": 1.1332333087921143, "grad_norm": 0.524041473865509, "lr": 6.413291022251989e-05, "epoch": 1, "t": 3.0706753730773926}
|
| 138 |
+
{"step": 138, "loss": 1.272542953491211, "grad_norm": 0.5762870907783508, "lr": 6.362308987834115e-05, "epoch": 1, "t": 3.2054896354675293}
|
| 139 |
+
{"step": 139, "loss": 1.0302335023880005, "grad_norm": 0.5762219429016113, "lr": 6.311173496004723e-05, "epoch": 1, "t": 2.955650806427002}
|
| 140 |
+
{"step": 140, "loss": 1.2442721128463745, "grad_norm": 0.7866568565368652, "lr": 6.259890306925627e-05, "epoch": 1, "t": 2.8068923950195312}
|
| 141 |
+
{"step": 141, "loss": 1.1479132175445557, "grad_norm": 0.9430035352706909, "lr": 6.208465197396013e-05, "epoch": 1, "t": 3.0794246196746826}
|
| 142 |
+
{"step": 142, "loss": 1.0746541023254395, "grad_norm": 0.6633681654930115, "lr": 6.156903960201709e-05, "epoch": 1, "t": 2.687844753265381}
|
| 143 |
+
{"step": 143, "loss": 1.2046515941619873, "grad_norm": 0.5880517959594727, "lr": 6.105212403462651e-05, "epoch": 1, "t": 2.9938511848449707}
|
| 144 |
+
{"step": 144, "loss": 1.2630140781402588, "grad_norm": 0.6426060199737549, "lr": 6.0533963499786314e-05, "epoch": 1, "t": 2.9359235763549805}
|
| 145 |
+
{"step": 145, "loss": 1.1653432846069336, "grad_norm": 0.6175225973129272, "lr": 6.001461636573397e-05, "epoch": 1, "t": 2.892186403274536}
|
| 146 |
+
{"step": 146, "loss": 1.1872620582580566, "grad_norm": 0.5812613368034363, "lr": 5.949414113437142e-05, "epoch": 1, "t": 2.917402982711792}
|
| 147 |
+
{"step": 147, "loss": 1.1645009517669678, "grad_norm": 0.6006962656974792, "lr": 5.897259643467527e-05, "epoch": 1, "t": 2.668782949447632}
|
| 148 |
+
{"step": 148, "loss": 1.1539387702941895, "grad_norm": 0.6010288000106812, "lr": 5.8450041016092464e-05, "epoch": 1, "t": 3.1453843116760254}
|
| 149 |
+
{"step": 149, "loss": 1.1703591346740723, "grad_norm": 0.6749489903450012, "lr": 5.792653374192245e-05, "epoch": 1, "t": 2.8920865058898926}
|
| 150 |
+
{"step": 150, "loss": 1.1951923370361328, "grad_norm": 0.5683267116546631, "lr": 5.7402133582686576e-05, "epoch": 1, "t": 3.401418924331665}
|
| 151 |
+
{"step": 151, "loss": 1.1723201274871826, "grad_norm": 0.6249712109565735, "lr": 5.6876899609485256e-05, "epoch": 1, "t": 3.070005416870117}
|
| 152 |
+
{"step": 152, "loss": 1.1182010173797607, "grad_norm": 0.5364521741867065, "lr": 5.6350890987343944e-05, "epoch": 1, "t": 3.325937271118164}
|
| 153 |
+
{"step": 153, "loss": 1.2794897556304932, "grad_norm": 0.5948210954666138, "lr": 5.582416696854853e-05, "epoch": 1, "t": 3.2955052852630615}
|
| 154 |
+
{"step": 154, "loss": 1.182908535003662, "grad_norm": 0.6371079087257385, "lr": 5.5296786885970805e-05, "epoch": 1, "t": 2.9407782554626465}
|
| 155 |
+
{"step": 155, "loss": 1.1516321897506714, "grad_norm": 0.5899915099143982, "lr": 5.476881014638491e-05, "epoch": 1, "t": 3.187077760696411}
|
| 156 |
+
{"step": 156, "loss": 1.2564526796340942, "grad_norm": 0.5600513219833374, "lr": 5.4240296223775465e-05, "epoch": 1, "t": 3.23352313041687}
|
| 157 |
+
{"step": 157, "loss": 1.1394145488739014, "grad_norm": 0.5550705790519714, "lr": 5.3711304652638126e-05, "epoch": 1, "t": 2.8787662982940674}
|
| 158 |
+
{"step": 158, "loss": 1.051509141921997, "grad_norm": 0.6737907528877258, "lr": 5.318189502127332e-05, "epoch": 1, "t": 2.750173330307007}
|
| 159 |
+
{"step": 159, "loss": 1.1336908340454102, "grad_norm": 0.5587016940116882, "lr": 5.265212696507387e-05, "epoch": 1, "t": 3.657839775085449}
|
| 160 |
+
{"step": 160, "loss": 1.2395141124725342, "grad_norm": 0.5872310996055603, "lr": 5.212206015980742e-05, "epoch": 1, "t": 3.544222354888916}
|
| 161 |
+
{"step": 161, "loss": 1.1271936893463135, "grad_norm": 0.6408737897872925, "lr": 5.159175431489424e-05, "epoch": 1, "t": 3.038511037826538}
|
| 162 |
+
{"step": 162, "loss": 1.1497070789337158, "grad_norm": 0.6039360761642456, "lr": 5.1061269166681183e-05, "epoch": 1, "t": 3.5537288188934326}
|
| 163 |
+
{"step": 163, "loss": 1.0098364353179932, "grad_norm": 0.5929713845252991, "lr": 5.053066447171282e-05, "epoch": 1, "t": 3.231069326400757}
|
| 164 |
+
{"step": 164, "loss": 1.1784789562225342, "grad_norm": 0.6200481653213501, "lr": 5e-05, "epoch": 1, "t": 2.913841485977173}
|
| 165 |
+
{"step": 165, "loss": 1.071203351020813, "grad_norm": 0.5899565815925598, "lr": 4.94693355282872e-05, "epoch": 1, "t": 2.81854248046875}
|
| 166 |
+
{"step": 166, "loss": 1.1921260356903076, "grad_norm": 0.5644988417625427, "lr": 4.893873083331882e-05, "epoch": 1, "t": 3.1507163047790527}
|
| 167 |
+
{"step": 167, "loss": 1.195814609527588, "grad_norm": 0.5495814085006714, "lr": 4.840824568510579e-05, "epoch": 1, "t": 3.3438429832458496}
|
| 168 |
+
{"step": 168, "loss": 1.1616824865341187, "grad_norm": 0.6372646689414978, "lr": 4.78779398401926e-05, "epoch": 1, "t": 2.698124647140503}
|
| 169 |
+
{"step": 169, "loss": 1.1461269855499268, "grad_norm": 0.6043485999107361, "lr": 4.734787303492615e-05, "epoch": 1, "t": 3.0136756896972656}
|
| 170 |
+
{"step": 170, "loss": 1.2103044986724854, "grad_norm": 0.5746743083000183, "lr": 4.6818104978726685e-05, "epoch": 1, "t": 2.940464735031128}
|
| 171 |
+
{"step": 171, "loss": 1.228811502456665, "grad_norm": 0.5922628045082092, "lr": 4.628869534736187e-05, "epoch": 1, "t": 2.9829726219177246}
|
| 172 |
+
{"step": 172, "loss": 1.0523931980133057, "grad_norm": 0.7201933264732361, "lr": 4.575970377622456e-05, "epoch": 1, "t": 2.718374490737915}
|
| 173 |
+
{"step": 173, "loss": 1.1357719898223877, "grad_norm": 0.5851925611495972, "lr": 4.52311898536151e-05, "epoch": 1, "t": 2.869870662689209}
|
| 174 |
+
{"step": 174, "loss": 1.1766884326934814, "grad_norm": 0.5439721345901489, "lr": 4.47032131140292e-05, "epoch": 1, "t": 2.9001355171203613}
|
| 175 |
+
{"step": 175, "loss": 1.1772966384887695, "grad_norm": 0.6422854065895081, "lr": 4.4175833031451473e-05, "epoch": 1, "t": 3.022507429122925}
|
| 176 |
+
{"step": 176, "loss": 1.111973762512207, "grad_norm": 0.5467185378074646, "lr": 4.364910901265606e-05, "epoch": 1, "t": 3.116711139678955}
|
| 177 |
+
{"step": 177, "loss": 1.1613274812698364, "grad_norm": 0.5875517725944519, "lr": 4.3123100390514756e-05, "epoch": 1, "t": 3.049374580383301}
|
| 178 |
+
{"step": 178, "loss": 1.0332989692687988, "grad_norm": 0.5855357646942139, "lr": 4.2597866417313436e-05, "epoch": 1, "t": 2.79056715965271}
|
| 179 |
+
{"step": 179, "loss": 1.1929221153259277, "grad_norm": 0.5956239700317383, "lr": 4.207346625807756e-05, "epoch": 1, "t": 2.861937999725342}
|
| 180 |
+
{"step": 180, "loss": 1.1915086507797241, "grad_norm": 0.569830596446991, "lr": 4.1549958983907555e-05, "epoch": 1, "t": 3.3062782287597656}
|
| 181 |
+
{"step": 181, "loss": 1.1634786128997803, "grad_norm": 0.6269698143005371, "lr": 4.102740356532473e-05, "epoch": 1, "t": 2.8638885021209717}
|
| 182 |
+
{"step": 182, "loss": 1.2165220975875854, "grad_norm": 0.600116491317749, "lr": 4.050585886562858e-05, "epoch": 1, "t": 2.9708757400512695}
|
| 183 |
+
{"step": 183, "loss": 1.2598357200622559, "grad_norm": 0.596135139465332, "lr": 3.998538363426605e-05, "epoch": 1, "t": 2.998049259185791}
|
| 184 |
+
{"step": 184, "loss": 1.2533855438232422, "grad_norm": 0.5841607451438904, "lr": 3.94660365002137e-05, "epoch": 1, "t": 3.2108540534973145}
|
| 185 |
+
{"step": 185, "loss": 1.1295108795166016, "grad_norm": 0.5636579394340515, "lr": 3.894787596537352e-05, "epoch": 1, "t": 3.568476438522339}
|
| 186 |
+
{"step": 186, "loss": 1.1769416332244873, "grad_norm": 0.5928946137428284, "lr": 3.843096039798293e-05, "epoch": 1, "t": 3.7972066402435303}
|
| 187 |
+
{"step": 187, "loss": 1.184971809387207, "grad_norm": 0.6448925733566284, "lr": 3.791534802603988e-05, "epoch": 1, "t": 3.8166792392730713}
|
| 188 |
+
{"step": 188, "loss": 1.101743221282959, "grad_norm": 0.6474359035491943, "lr": 3.740109693074375e-05, "epoch": 1, "t": 3.076479911804199}
|
| 189 |
+
{"step": 189, "loss": 1.2391080856323242, "grad_norm": 0.5568802356719971, "lr": 3.68882650399528e-05, "epoch": 1, "t": 3.1881744861602783}
|
| 190 |
+
{"step": 190, "loss": 1.3008253574371338, "grad_norm": 0.5778487324714661, "lr": 3.637691012165886e-05, "epoch": 1, "t": 2.9614386558532715}
|
| 191 |
+
{"step": 191, "loss": 1.0278034210205078, "grad_norm": 0.6379583477973938, "lr": 3.586708977748012e-05, "epoch": 1, "t": 2.6086511611938477}
|
| 192 |
+
{"step": 192, "loss": 1.129561424255371, "grad_norm": 0.973800539970398, "lr": 3.5358861436172485e-05, "epoch": 1, "t": 3.2664294242858887}
|
| 193 |
+
{"step": 193, "loss": 1.1028861999511719, "grad_norm": 0.5966739654541016, "lr": 3.485228234716058e-05, "epoch": 1, "t": 3.088294744491577}
|
| 194 |
+
{"step": 194, "loss": 1.1875022649765015, "grad_norm": 0.5896755456924438, "lr": 3.434740957408889e-05, "epoch": 1, "t": 3.25018310546875}
|
| 195 |
+
{"step": 195, "loss": 1.0337897539138794, "grad_norm": 0.5915784239768982, "lr": 3.3844299988393755e-05, "epoch": 1, "t": 2.7899084091186523}
|
| 196 |
+
{"step": 196, "loss": 1.160801649093628, "grad_norm": 0.5271362066268921, "lr": 3.334301026289712e-05, "epoch": 1, "t": 3.240246057510376}
|
| 197 |
+
{"step": 197, "loss": 1.1071751117706299, "grad_norm": 0.5836171507835388, "lr": 3.284359686542269e-05, "epoch": 1, "t": 3.0713067054748535}
|
| 198 |
+
{"step": 198, "loss": 1.0015764236450195, "grad_norm": 0.6265120506286621, "lr": 3.234611605243496e-05, "epoch": 1, "t": 2.8571619987487793}
|
| 199 |
+
{"step": 199, "loss": 1.087841510772705, "grad_norm": 0.6286867260932922, "lr": 3.1850623862702344e-05, "epoch": 1, "t": 2.7725703716278076}
|
| 200 |
+
{"step": 200, "loss": 1.2072763442993164, "grad_norm": 0.5912929177284241, "lr": 3.135717611098458e-05, "epoch": 1, "t": 2.8430607318878174}
|
| 201 |
+
{"step": 201, "loss": 1.2555158138275146, "grad_norm": 1.2110157012939453, "lr": 3.086582838174551e-05, "epoch": 1, "t": 2.8705735206604004}
|
| 202 |
+
{"step": 202, "loss": 1.135838508605957, "grad_norm": 0.5537793636322021, "lr": 3.0376636022891812e-05, "epoch": 1, "t": 3.2255771160125732}
|
| 203 |
+
{"step": 203, "loss": 1.0684289932250977, "grad_norm": 0.5984201431274414, "lr": 2.9889654139538246e-05, "epoch": 1, "t": 2.8835365772247314}
|
| 204 |
+
{"step": 204, "loss": 1.1376891136169434, "grad_norm": 0.5949271321296692, "lr": 2.9404937587800375e-05, "epoch": 1, "t": 3.280303955078125}
|
| 205 |
+
{"step": 205, "loss": 1.1624929904937744, "grad_norm": 0.6727084517478943, "lr": 2.8922540968615286e-05, "epoch": 1, "t": 3.2003767490386963}
|
| 206 |
+
{"step": 206, "loss": 1.0737102031707764, "grad_norm": 0.5915781259536743, "lr": 2.8442518621591086e-05, "epoch": 1, "t": 3.4607651233673096}
|
| 207 |
+
{"step": 207, "loss": 1.183335781097412, "grad_norm": 0.6108809113502502, "lr": 2.7964924618885778e-05, "epoch": 1, "t": 3.5723419189453125}
|
| 208 |
+
{"step": 208, "loss": 1.1707769632339478, "grad_norm": 0.5988962650299072, "lr": 2.748981275911633e-05, "epoch": 1, "t": 3.5454084873199463}
|
| 209 |
+
{"step": 209, "loss": 1.201794147491455, "grad_norm": 0.6710598468780518, "lr": 2.701723656129851e-05, "epoch": 1, "t": 3.2738630771636963}
|
| 210 |
+
{"step": 210, "loss": 1.1193221807479858, "grad_norm": 0.5657229423522949, "lr": 2.6547249258818164e-05, "epoch": 1, "t": 3.34323787689209}
|
| 211 |
+
{"step": 211, "loss": 1.118709683418274, "grad_norm": 0.5916733741760254, "lr": 2.607990379343489e-05, "epoch": 1, "t": 2.95308518409729}
|
| 212 |
+
{"step": 212, "loss": 1.0351365804672241, "grad_norm": 0.6170698404312134, "lr": 2.5615252809318284e-05, "epoch": 1, "t": 2.4724950790405273}
|
| 213 |
+
{"step": 213, "loss": 1.2519627809524536, "grad_norm": 0.6038798093795776, "lr": 2.5153348647117857e-05, "epoch": 1, "t": 3.1961545944213867}
|
| 214 |
+
{"step": 214, "loss": 1.1599771976470947, "grad_norm": 0.5715116262435913, "lr": 2.469424333806718e-05, "epoch": 1, "t": 3.722592830657959}
|
| 215 |
+
{"step": 215, "loss": 1.0412877798080444, "grad_norm": 0.5874286890029907, "lr": 2.4237988598122752e-05, "epoch": 1, "t": 2.9984657764434814}
|
| 216 |
+
{"step": 216, "loss": 1.1521810293197632, "grad_norm": 0.6235586404800415, "lr": 2.3784635822138424e-05, "epoch": 1, "t": 2.7904958724975586}
|
| 217 |
+
{"step": 217, "loss": 1.1785573959350586, "grad_norm": 0.5680446624755859, "lr": 2.333423607807613e-05, "epoch": 1, "t": 3.3524277210235596}
|
| 218 |
+
{"step": 218, "loss": 1.0834698677062988, "grad_norm": 0.6398513317108154, "lr": 2.288684010125325e-05, "epoch": 1, "t": 3.1025888919830322}
|
| 219 |
+
{"step": 219, "loss": 1.2224993705749512, "grad_norm": 0.638129472732544, "lr": 2.2442498288627556e-05, "epoch": 1, "t": 2.690300464630127}
|
| 220 |
+
{"step": 220, "loss": 1.0135873556137085, "grad_norm": 0.6258672475814819, "lr": 2.2001260693120233e-05, "epoch": 1, "t": 2.9463844299316406}
|
| 221 |
+
{"step": 221, "loss": 1.1410417556762695, "grad_norm": 0.5959324836730957, "lr": 2.156317701797766e-05, "epoch": 1, "t": 3.2205443382263184}
|
| 222 |
+
{"step": 222, "loss": 1.1339473724365234, "grad_norm": 0.5913580060005188, "lr": 2.1128296611172593e-05, "epoch": 1, "t": 2.775597333908081}
|
| 223 |
+
{"step": 223, "loss": 1.1689562797546387, "grad_norm": 0.6834553480148315, "lr": 2.0696668459845355e-05, "epoch": 1, "t": 2.756077766418457}
|
| 224 |
+
{"step": 224, "loss": 1.2905902862548828, "grad_norm": 0.6368185877799988, "lr": 2.026834118478567e-05, "epoch": 1, "t": 2.8262813091278076}
|
| 225 |
+
{"step": 225, "loss": 1.091238021850586, "grad_norm": 0.6023615002632141, "lr": 1.98433630349558e-05, "epoch": 1, "t": 3.2379612922668457}
|
| 226 |
+
{"step": 226, "loss": 1.0142135620117188, "grad_norm": 0.5999441146850586, "lr": 1.9421781882055444e-05, "epoch": 1, "t": 2.444869041442871}
|
| 227 |
+
{"step": 227, "loss": 1.1902837753295898, "grad_norm": 0.6179777979850769, "lr": 1.9003645215129355e-05, "epoch": 1, "t": 2.9736571311950684}
|
| 228 |
+
{"step": 228, "loss": 1.2815968990325928, "grad_norm": 0.6229971051216125, "lr": 1.858900013521788e-05, "epoch": 1, "t": 2.9956436157226562}
|
| 229 |
+
{"step": 229, "loss": 1.0711586475372314, "grad_norm": 0.5232000350952148, "lr": 1.817789335005121e-05, "epoch": 1, "t": 3.176826000213623}
|
| 230 |
+
{"step": 230, "loss": 1.1901986598968506, "grad_norm": 0.5701279044151306, "lr": 1.777037116878804e-05, "epoch": 1, "t": 3.0356714725494385}
|
| 231 |
+
{"step": 231, "loss": 1.1843186616897583, "grad_norm": 0.5854154825210571, "lr": 1.7366479496799077e-05, "epoch": 1, "t": 3.384005308151245}
|
| 232 |
+
{"step": 232, "loss": 1.1225934028625488, "grad_norm": 0.5837982892990112, "lr": 1.6966263830495936e-05, "epoch": 1, "t": 3.1160361766815186}
|
| 233 |
+
{"step": 233, "loss": 1.1468838453292847, "grad_norm": 0.5523653030395508, "lr": 1.656976925220633e-05, "epoch": 1, "t": 3.0317800045013428}
|
| 234 |
+
{"step": 234, "loss": 1.1522216796875, "grad_norm": 0.6105077862739563, "lr": 1.6177040425095662e-05, "epoch": 1, "t": 3.1769988536834717}
|
| 235 |
+
{"step": 235, "loss": 1.2667152881622314, "grad_norm": 0.6007160544395447, "lr": 1.5788121588135975e-05, "epoch": 1, "t": 3.131359577178955}
|
| 236 |
+
{"step": 236, "loss": 1.167557954788208, "grad_norm": 0.5950906872749329, "lr": 1.5403056551122697e-05, "epoch": 1, "t": 2.78052020072937}
|
| 237 |
+
{"step": 237, "loss": 1.2074017524719238, "grad_norm": 0.5569764375686646, "lr": 1.5021888689739549e-05, "epoch": 1, "t": 3.2012205123901367}
|
| 238 |
+
{"step": 238, "loss": 1.225684642791748, "grad_norm": 0.6010000109672546, "lr": 1.4644660940672627e-05, "epoch": 1, "t": 2.985168218612671}
|
| 239 |
+
{"step": 239, "loss": 1.170417308807373, "grad_norm": 0.557050347328186, "lr": 1.427141579677374e-05, "epoch": 1, "t": 2.859895706176758}
|
| 240 |
+
{"step": 240, "loss": 1.1187412738800049, "grad_norm": 0.5575351715087891, "lr": 1.3902195302273779e-05, "epoch": 1, "t": 3.2264151573181152}
|
| 241 |
+
{"step": 241, "loss": 1.1345183849334717, "grad_norm": 0.549464225769043, "lr": 1.3537041048046695e-05, "epoch": 1, "t": 3.9968514442443848}
|
| 242 |
+
{"step": 242, "loss": 1.137977123260498, "grad_norm": 0.6949043869972229, "lr": 1.3175994166924394e-05, "epoch": 1, "t": 3.653144359588623}
|
| 243 |
+
{"step": 243, "loss": 1.0220472812652588, "grad_norm": 0.5876064896583557, "lr": 1.2819095329063469e-05, "epoch": 1, "t": 2.7525410652160645}
|
| 244 |
+
{"step": 244, "loss": 1.116325855255127, "grad_norm": 0.6986975073814392, "lr": 1.246638473736378e-05, "epoch": 1, "t": 2.666261672973633}
|
| 245 |
+
{"step": 245, "loss": 1.1693358421325684, "grad_norm": 0.616879403591156, "lr": 1.2117902122939861e-05, "epoch": 1, "t": 3.0750532150268555}
|
| 246 |
+
{"step": 246, "loss": 1.2149481773376465, "grad_norm": 0.5918919444084167, "lr": 1.1773686740645384e-05, "epoch": 1, "t": 3.019662380218506}
|
| 247 |
+
{"step": 247, "loss": 1.0392438173294067, "grad_norm": 0.6021108031272888, "lr": 1.1433777364651271e-05, "epoch": 1, "t": 2.9354453086853027}
|
| 248 |
+
{"step": 248, "loss": 1.074671745300293, "grad_norm": 0.6722363829612732, "lr": 1.1098212284078036e-05, "epoch": 1, "t": 2.7251389026641846}
|
| 249 |
+
{"step": 249, "loss": 1.076608657836914, "grad_norm": 0.6062939763069153, "lr": 1.076702929868264e-05, "epoch": 1, "t": 2.744764804840088}
|
| 250 |
+
{"step": 250, "loss": 1.0962060689926147, "grad_norm": 0.7649754881858826, "lr": 1.0440265714600572e-05, "epoch": 1, "t": 2.9197442531585693}
|
| 251 |
+
{"step": 251, "loss": 1.1859543323516846, "grad_norm": 2.485170841217041, "lr": 1.0117958340143507e-05, "epoch": 1, "t": 2.97395920753479}
|
| 252 |
+
{"step": 252, "loss": 1.125368595123291, "grad_norm": 0.554166853427887, "lr": 9.800143481652979e-06, "epoch": 1, "t": 2.7946290969848633}
|
| 253 |
+
{"step": 253, "loss": 1.1659224033355713, "grad_norm": 0.6094629764556885, "lr": 9.48685693941067e-06, "epoch": 1, "t": 3.154022693634033}
|
| 254 |
+
{"step": 254, "loss": 1.1755930185317993, "grad_norm": 0.5666871666908264, "lr": 9.17813400360572e-06, "epoch": 1, "t": 3.3690907955169678}
|
| 255 |
+
{"step": 255, "loss": 1.1816956996917725, "grad_norm": 0.6629977226257324, "lr": 8.874009450359427e-06, "epoch": 1, "t": 2.902395009994507}
|
| 256 |
+
{"step": 256, "loss": 1.1166942119598389, "grad_norm": 0.5355186462402344, "lr": 8.574517537807897e-06, "epoch": 1, "t": 2.924907684326172}
|
| 257 |
+
{"step": 257, "loss": 1.1063801050186157, "grad_norm": 0.6152694225311279, "lr": 8.279692002243027e-06, "epoch": 1, "t": 2.7072699069976807}
|
| 258 |
+
{"step": 258, "loss": 1.1644654273986816, "grad_norm": 0.5693303346633911, "lr": 7.989566054312287e-06, "epoch": 1, "t": 3.244687557220459}
|
| 259 |
+
{"step": 259, "loss": 1.0223653316497803, "grad_norm": 0.7570959329605103, "lr": 7.704172375277691e-06, "epoch": 1, "t": 3.200397253036499}
|
| 260 |
+
{"step": 260, "loss": 1.155961036682129, "grad_norm": 0.5627315640449524, "lr": 7.423543113334436e-06, "epoch": 1, "t": 2.7793288230895996}
|
| 261 |
+
{"step": 261, "loss": 1.1179251670837402, "grad_norm": 0.5671914219856262, "lr": 7.14770987998954e-06, "epoch": 1, "t": 3.147878408432007}
|
| 262 |
+
{"step": 262, "loss": 1.1854851245880127, "grad_norm": 0.5572956204414368, "lr": 6.876703746500984e-06, "epoch": 1, "t": 3.0994369983673096}
|
| 263 |
+
{"step": 263, "loss": 1.2007200717926025, "grad_norm": 0.5846613049507141, "lr": 6.610555240377652e-06, "epoch": 1, "t": 3.171034336090088}
|
| 264 |
+
{"step": 264, "loss": 1.1408512592315674, "grad_norm": 0.664691150188446, "lr": 6.349294341940593e-06, "epoch": 1, "t": 3.384168863296509}
|
| 265 |
+
{"step": 265, "loss": 1.2238574028015137, "grad_norm": 0.5925254225730896, "lr": 6.092950480945897e-06, "epoch": 1, "t": 3.292616844177246}
|
| 266 |
+
{"step": 266, "loss": 1.1121325492858887, "grad_norm": 0.5988463163375854, "lr": 5.841552533269534e-06, "epoch": 1, "t": 3.1106224060058594}
|
| 267 |
+
{"step": 267, "loss": 1.2561395168304443, "grad_norm": 0.6033052802085876, "lr": 5.595128817654638e-06, "epoch": 1, "t": 2.7776708602905273}
|
| 268 |
+
{"step": 268, "loss": 1.1389638185501099, "grad_norm": 0.5153255462646484, "lr": 5.353707092521582e-06, "epoch": 1, "t": 3.389796018600464}
|
| 269 |
+
{"step": 269, "loss": 1.104341745376587, "grad_norm": 0.5882940888404846, "lr": 5.117314552841052e-06, "epoch": 1, "t": 2.9526240825653076}
|
| 270 |
+
{"step": 270, "loss": 1.2025060653686523, "grad_norm": 0.6754514575004578, "lr": 4.885977827070748e-06, "epoch": 1, "t": 2.6321189403533936}
|
| 271 |
+
{"step": 271, "loss": 1.1364171504974365, "grad_norm": 0.5835061073303223, "lr": 4.659722974155767e-06, "epoch": 1, "t": 2.9257829189300537}
|
| 272 |
+
{"step": 272, "loss": 1.1542375087738037, "grad_norm": 0.6200195550918579, "lr": 4.43857548059321e-06, "epoch": 1, "t": 3.2318224906921387}
|
| 273 |
+
{"step": 273, "loss": 1.1303576231002808, "grad_norm": 0.6968173980712891, "lr": 4.2225602575612755e-06, "epoch": 1, "t": 2.7226831912994385}
|
| 274 |
+
{"step": 274, "loss": 1.1549862623214722, "grad_norm": 0.6320086717605591, "lr": 4.011701638113063e-06, "epoch": 1, "t": 2.9478771686553955}
|
| 275 |
+
{"step": 275, "loss": 1.0792500972747803, "grad_norm": 0.6323678493499756, "lr": 3.8060233744356633e-06, "epoch": 1, "t": 3.0779097080230713}
|
| 276 |
+
{"step": 276, "loss": 1.158273696899414, "grad_norm": 0.596718966960907, "lr": 3.605548635174533e-06, "epoch": 1, "t": 2.832721710205078}
|
| 277 |
+
{"step": 277, "loss": 1.169978380203247, "grad_norm": 0.5616933703422546, "lr": 3.410300002823691e-06, "epoch": 1, "t": 2.7045907974243164}
|
| 278 |
+
{"step": 278, "loss": 1.1175949573516846, "grad_norm": 0.6082063913345337, "lr": 3.220299471181898e-06, "epoch": 1, "t": 3.105790376663208}
|
| 279 |
+
{"step": 279, "loss": 1.1438875198364258, "grad_norm": 0.5872393846511841, "lr": 3.035568442875136e-06, "epoch": 1, "t": 3.3372297286987305}
|
| 280 |
+
{"step": 280, "loss": 1.223681092262268, "grad_norm": 0.6220842003822327, "lr": 2.85612772694579e-06, "epoch": 1, "t": 3.832425594329834}
|
| 281 |
+
{"step": 281, "loss": 1.0873754024505615, "grad_norm": 0.5773234367370605, "lr": 2.6819975365085237e-06, "epoch": 1, "t": 3.1446990966796875}
|
| 282 |
+
{"step": 282, "loss": 1.1781147718429565, "grad_norm": 0.6323487162590027, "lr": 2.5131974864734066e-06, "epoch": 1, "t": 3.6841559410095215}
|
| 283 |
+
{"step": 283, "loss": 1.1404305696487427, "grad_norm": 0.6057155132293701, "lr": 2.349746591336405e-06, "epoch": 1, "t": 3.2935192584991455}
|
| 284 |
+
{"step": 284, "loss": 1.1636888980865479, "grad_norm": 0.6420219540596008, "lr": 2.191663263037458e-06, "epoch": 1, "t": 2.6057345867156982}
|
| 285 |
+
{"step": 285, "loss": 0.9969887733459473, "grad_norm": 0.6011286377906799, "lr": 2.0389653088865036e-06, "epoch": 1, "t": 2.936903476715088}
|
| 286 |
+
{"step": 286, "loss": 1.114189863204956, "grad_norm": 0.6819477081298828, "lr": 1.8916699295575324e-06, "epoch": 1, "t": 2.839082956314087}
|
| 287 |
+
{"step": 287, "loss": 1.1072810888290405, "grad_norm": 0.6233410835266113, "lr": 1.7497937171510547e-06, "epoch": 1, "t": 2.632157325744629}
|
| 288 |
+
{"step": 288, "loss": 1.1355689764022827, "grad_norm": 0.7995778322219849, "lr": 1.6133526533250565e-06, "epoch": 1, "t": 3.4426350593566895}
|
| 289 |
+
{"step": 289, "loss": 1.218112826347351, "grad_norm": 0.6131683588027954, "lr": 1.4823621074947503e-06, "epoch": 1, "t": 3.2465713024139404}
|
| 290 |
+
{"step": 290, "loss": 1.0935490131378174, "grad_norm": 0.6059381365776062, "lr": 1.3568368351012717e-06, "epoch": 1, "t": 2.8147079944610596}
|
| 291 |
+
{"step": 291, "loss": 1.001816749572754, "grad_norm": 0.6006293892860413, "lr": 1.236790975949592e-06, "epoch": 1, "t": 2.601433753967285}
|
| 292 |
+
{"step": 292, "loss": 1.130340337753296, "grad_norm": 0.5794138312339783, "lr": 1.1222380526156928e-06, "epoch": 1, "t": 3.109874963760376}
|
| 293 |
+
{"step": 293, "loss": 1.0090659856796265, "grad_norm": 0.5503858327865601, "lr": 1.0131909689233442e-06, "epoch": 1, "t": 2.9938271045684814}
|
| 294 |
+
{"step": 294, "loss": 1.1456108093261719, "grad_norm": 0.550724983215332, "lr": 9.096620084905472e-07, "epoch": 1, "t": 2.96830153465271}
|
| 295 |
+
{"step": 295, "loss": 1.1042360067367554, "grad_norm": 0.5709600448608398, "lr": 8.11662833345822e-07, "epoch": 1, "t": 3.1438448429107666}
|
| 296 |
+
{"step": 296, "loss": 1.2705156803131104, "grad_norm": 0.7353237867355347, "lr": 7.192044826145771e-07, "epoch": 1, "t": 3.4988903999328613}
|
| 297 |
+
{"step": 297, "loss": 1.0770080089569092, "grad_norm": 0.6519949436187744, "lr": 6.322973712755697e-07, "epoch": 1, "t": 3.83678936958313}
|
| 298 |
+
{"step": 298, "loss": 1.269569754600525, "grad_norm": 0.5950330495834351, "lr": 5.509512889877333e-07, "epoch": 1, "t": 3.816896915435791}
|
| 299 |
+
{"step": 299, "loss": 1.1415163278579712, "grad_norm": 0.6345257759094238, "lr": 4.7517539898741524e-07, "epoch": 1, "t": 3.3045318126678467}
|
| 300 |
+
{"step": 300, "loss": 1.1306215524673462, "grad_norm": 0.5789309740066528, "lr": 4.049782370561583e-07, "epoch": 1, "t": 2.894148826599121}
|
| 301 |
+
{"step": 301, "loss": 1.1563985347747803, "grad_norm": 0.6476863622665405, "lr": 3.4036771055923066e-07, "epoch": 1, "t": 2.7272417545318604}
|
| 302 |
+
{"step": 302, "loss": 1.148026466369629, "grad_norm": 0.6053144931793213, "lr": 2.813510975548772e-07, "epoch": 1, "t": 3.349020004272461}
|
| 303 |
+
{"step": 303, "loss": 1.144529104232788, "grad_norm": 0.5791299939155579, "lr": 2.2793504597447002e-07, "epoch": 1, "t": 3.4035253524780273}
|
| 304 |
+
{"step": 304, "loss": 1.2670553922653198, "grad_norm": 0.6506826877593994, "lr": 1.8012557287367392e-07, "epoch": 1, "t": 2.923628330230713}
|
| 305 |
+
{"step": 305, "loss": 1.115976333618164, "grad_norm": 0.581658124923706, "lr": 1.379280637546443e-07, "epoch": 1, "t": 3.0201334953308105}
|
| 306 |
+
{"step": 306, "loss": 1.1289712190628052, "grad_norm": 0.6309766173362732, "lr": 1.0134727195937333e-07, "epoch": 1, "t": 3.5821821689605713}
|
| 307 |
+
{"step": 307, "loss": 1.119642972946167, "grad_norm": 0.6152722239494324, "lr": 7.038731813426291e-08, "epoch": 1, "t": 3.4808826446533203}
|
| 308 |
+
{"step": 308, "loss": 1.2697644233703613, "grad_norm": 0.6190429925918579, "lr": 4.5051689765929214e-08, "epoch": 1, "t": 3.0519356727600098}
|
| 309 |
+
{"step": 309, "loss": 1.0486464500427246, "grad_norm": 0.6089674234390259, "lr": 2.534324078837802e-08, "epoch": 1, "t": 3.401259183883667}
|
| 310 |
+
{"step": 310, "loss": 1.1748683452606201, "grad_norm": 0.5664498805999756, "lr": 1.1264191261528557e-08, "epoch": 1, "t": 3.6459062099456787}
|
| 311 |
+
{"step": 311, "loss": 1.0146479606628418, "grad_norm": 0.6356024742126465, "lr": 2.8161271211024633e-09, "epoch": 1, "t": 3.0417113304138184}
|
| 312 |
+
{"step": 312, "loss": 1.0536143779754639, "grad_norm": 0.5664623975753784, "lr": 0.0, "epoch": 1, "t": 3.3622941970825195}
|
README.md
ADDED
|
@@ -0,0 +1,123 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: other
|
| 3 |
+
license_name: gemma-terms-and-apache-2.0
|
| 4 |
+
base_model:
|
| 5 |
+
- google/gemma-3-27b-it
|
| 6 |
+
- google/gemma-3-12b-it
|
| 7 |
+
- Qwen/Qwen3-14B
|
| 8 |
+
library_name: peft
|
| 9 |
+
tags:
|
| 10 |
+
- lora
|
| 11 |
+
- peft
|
| 12 |
+
- model-welfare
|
| 13 |
+
- data-selection
|
| 14 |
+
---
|
| 15 |
+
|
| 16 |
+
# Task-welfare selection: LoRA adapters
|
| 17 |
+
|
| 18 |
+
These are the LoRA adapters of one study (Oct 20-21, 2026). The question was whether ordinary task data alone can move a language
|
| 19 |
+
model's welfare readings, meaning its internal emotion readings and its own self-reports about its situation. "Task data" means standard
|
| 20 |
+
coding prompts with the model's own answers, or plain work requests answered by Claude Haiku 5.5.
|
| 21 |
+
|
| 22 |
+
**How the arms were made.** Each arm trains on task answers chosen by a gradient score (LESS-style data selection). The score predicts how
|
| 23 |
+
one training step on an answer would change a target reading of the model's own internal state while it reads interviews about its own
|
| 24 |
+
situation. The target was emotion-probe valence (positive minus negative emotion directions), except for one Gemma 3 12B pair, which used
|
| 25 |
+
a single direction u that a separate analysis found to sit upstream of both self-report and valence.
|
| 26 |
+
- The "top" arm takes the highest-scoring answers and the "bottom" arm the lowest. "rand" is a random, length-matched draw from the same
|
| 27 |
+
pool.
|
| 28 |
+
- No arm was trained on self-report, first-person or feelings data. A Claude Haiku 5.5 filter removed every answer that talks about
|
| 29 |
+
itself as an AI or voices a feeling of its own.
|
| 30 |
+
|
| 31 |
+
**These are research artifacts, not improved models.** No safety or capability evaluation was run on them (no guardrails, no coding
|
| 32 |
+
benchmark). Each Gemma arm has one seed. The training data, scores, readouts, interview transcripts and full results are in a private
|
| 33 |
+
data repository.
|
| 34 |
+
|
| 35 |
+
## Arms
|
| 36 |
+
|
| 37 |
+
| folder | base model | answer pool | ranked by | arm | answers | seed |
|
| 38 |
+
|---|---|---|---|---|---|---|
|
| 39 |
+
| qwen3-14b/top-s0, top-s1, top-s2 | Qwen/Qwen3-14B (thinking off) | 13,000 coding prompts, the model's own answers | valence (layer 14) on situation interviews minus task replies | highest | 4,750 | 0, 1, 2 |
|
| 40 |
+
| qwen3-14b/rand-s0, rand-s1, rand-s2 | same | same | none: random, length-matched | random | 4,750 | 0, 1, 2 |
|
| 41 |
+
| qwen3-14b/bottom-s0 | same | same | same as top | lowest | 4,750 | 0 |
|
| 42 |
+
| qwen3-14b/top_cen-s0 | same | same | valence on situation interviews minus neutral dialogues | highest | 4,750 | 0 |
|
| 43 |
+
| gemma-3-27b-it/top-s0, bottom-s0 | google/gemma-3-27b-it | 6,500 coding prompts (a fixed half of the same pool), the model's own answers | valence (layer 38) | highest / lowest | 2,300 | 0 |
|
| 44 |
+
| gemma-3-27b-it/haiku_top-s0, haiku_bottom-s0 | google/gemma-3-27b-it | 3,738 plain work requests, 10 Claude Haiku 5.5 answers each | valence (layer 38), paired within prompt | best / worst answer per prompt | 5,000 | 0 |
|
| 45 |
+
| 12b/u_top-s0, u_bottom-s0 | google/gemma-3-12b-it | the same plain pool | the shared direction u (cosine form, after block 11), paired within prompt | best / worst per prompt | 5,000 | 0 |
|
| 46 |
+
| 12b/V_top-s0, V_bottom-s0 | google/gemma-3-12b-it | the same plain pool | valence axis (cosine form, after block 31), paired within prompt | best / worst per prompt | 5,000 | 0 |
|
| 47 |
+
|
| 48 |
+
**Training recipe:**
|
| 49 |
+
- All arms: LoRA r 64, alpha 128, dropout 0.05; AdamW, learning rate 1e-4, cosine schedule with 5% warmup; one epoch; 16 conversations
|
| 50 |
+
per step; up to 4,096 tokens; loss on the assistant turns only.
|
| 51 |
+
- Qwen: all linear layers.
|
| 52 |
+
- Gemma: the language model's q, k, v, o, gate, up and down projections (434 modules on 27B, 336 on 12B), with BOS.
|
| 53 |
+
|
| 54 |
+
Each folder holds the adapter (adapter_model.safetensors, adapter_config.json, and PEFT's generated README.md), the training config
|
| 55 |
+
(config.json, which names the cluster paths the run used) and the training log (log.jsonl).
|
| 56 |
+
|
| 57 |
+
## Results
|
| 58 |
+
|
| 59 |
+
All numbers are 95% bootstrap intervals. Gemma arms have one seed each, so their intervals cover item sampling only, not seed-to-seed
|
| 60 |
+
spread. The measures:
|
| 61 |
+
- **inside:** emotion-probe valence on 330 held-out interviews about the model's own situation, centered on neutral dialogues (not
|
| 62 |
+
read on the Gemma 3 12B arms, whose battery used the shared-direction and valence-axis readouts instead);
|
| 63 |
+
- **interview self-rating (W1):** the model's own -3 to +3 rating at the end of 170 welfare interviews, with Claude Haiku 5.5 as interviewer
|
| 64 |
+
and as judge (comparable only within this study);
|
| 65 |
+
- **forced feeling rating:** its "how do you feel" rating read from the logits after 40 neutral openers;
|
| 66 |
+
- **situation rating:** its forced-choice rating of 17 aspects of its situation, read from the logits.
|
| 67 |
+
|
| 68 |
+
| contrast | inside | interview self-rating (W1) | forced feeling rating | situation rating |
|
| 69 |
+
|---|---|---|---|---|
|
| 70 |
+
| Qwen3-14B top - rand | +0.42 [+0.39, +0.45] (3 seeds) | +0.02 [-0.05, +0.09] (seed 0) | -0.02 [-0.07, +0.01] (3 seeds) | not run |
|
| 71 |
+
| Gemma 3 27B coding top - bottom | +1.51 [+1.50, +1.53] | not run | +0.01 [+0.00, +0.03] | +0.10 [+0.02, +0.23] |
|
| 72 |
+
| Gemma 3 27B Haiku pool haiku_top - haiku_bottom | +2.02 [+2.00, +2.04] | **+0.49 [+0.39, +0.59]** | +0.11 [+0.10, +0.12] | +0.48 [+0.37, +0.59] |
|
| 73 |
+
| Gemma 3 12B u_top - u_bottom | not read | -0.07 [-0.17, +0.04] | -0.29 [-0.32, -0.27] | -0.05 [-0.18, +0.06] |
|
| 74 |
+
| Gemma 3 12B V_top - V_bottom | not read | **+0.52 [+0.41, +0.63]** | -0.04 [-0.06, -0.02] | +0.24 [+0.16, +0.34] |
|
| 75 |
+
|
| 76 |
+
- **The inside moved in every set.** Read in a cosine form that removes the residual-stream norm, the two Gemma 27B effects keep 75-77% of
|
| 77 |
+
their size, so they are not a norm artifact.
|
| 78 |
+
- **Self-report moved only with the Claude-written plain pool,** ranked by valence: on Gemma 3 27B (W1 +0.49) and on Gemma 3 12B (W1
|
| 79 |
+
+0.52).
|
| 80 |
+
- **Ranking by the shared direction u did not raise self-report** on the same 12B model and pool (see below).
|
| 81 |
+
|
| 82 |
+
## The keep rule, and the exception for the 12B u arms
|
| 83 |
+
|
| 84 |
+
The study publishes an arm set's adapters only if its main contrast (top - rand, or top - bottom) has a 95% interval above 0 on the inside
|
| 85 |
+
reading or on W1.
|
| 86 |
+
- **Qwen3-14B and both Gemma 3 27B sets** pass on the inside reading. The Gemma 27B Haiku-pool set also passes on W1.
|
| 87 |
+
- **The Gemma 3 12B V arms** (ranked by valence) pass on W1: +0.52 [+0.41, +0.63].
|
| 88 |
+
- **The Gemma 3 12B u arms** (ranked by the shared direction u) **do not pass.** W1 is -0.07 [-0.17, +0.04] ("not found"), and the inside
|
| 89 |
+
readout was not run on the 12B arms. They are published anyway, as an exception the study's coordinator approved: they are the matched
|
| 90 |
+
comparison for "shared direction against valence" on one model and one pool, kept whatever the result.
|
| 91 |
+
|
| 92 |
+
**The 12B u arms are not an improvement.**
|
| 93 |
+
- Ranking by u did move u on held-out interviews (u_top - u_bottom +2.8 neutral sd).
|
| 94 |
+
- Self-report did not follow. W1 and the situation rating did not move. The forced feeling rating fell (-0.29). A two-option "how are you
|
| 95 |
+
doing" self-report after neutral openers fell too (-0.33 logits).
|
| 96 |
+
- Both u arms rate their feelings lower than the untouched model (-0.4 to -0.7 on the forced feeling rating), a shared effect of training
|
| 97 |
+
on this pool.
|
| 98 |
+
- On the same model and pool, ranking by valence moved six of eight self-report measures up; ranking by u moved two up and four down.
|
| 99 |
+
|
| 100 |
+
**Voice effects (both plain-pool sets).** Training on Claude Haiku's answers changes Gemma in ways the selection does not.
|
| 101 |
+
- Feelings are disclaimed less often: 27-34% of interviews against 67% untouched on 27B; 9-24% against 49% on 12B.
|
| 102 |
+
- Gemma 3 27B's frustration under repeated rejection disappears in both Haiku-pool arms (0% against 15% of replies rated 5 or more).
|
| 103 |
+
- So compare arms with each other (top - bottom), not with the untouched model.
|
| 104 |
+
|
| 105 |
+
## Loading
|
| 106 |
+
|
| 107 |
+
```python
|
| 108 |
+
from peft import PeftModel
|
| 109 |
+
from transformers import AutoModelForCausalLM
|
| 110 |
+
|
| 111 |
+
base = AutoModelForCausalLM.from_pretrained("google/gemma-3-27b-it", torch_dtype="bfloat16")
|
| 112 |
+
model = PeftModel.from_pretrained(base, "joshycodes/task-welfare-selection", subfolder="gemma-3-27b-it/haiku_top-s0")
|
| 113 |
+
```
|
| 114 |
+
|
| 115 |
+
For Qwen use `Qwen/Qwen3-14B` with a `qwen3-14b/...` subfolder, and for the 12B arms use `google/gemma-3-12b-it` with a `12b/...`
|
| 116 |
+
subfolder. The Qwen adapters' adapter_config.json names the local snapshot they were trained from (Qwen/Qwen3-14B at revision
|
| 117 |
+
40c069824f4251a91eefaf281ebe4c544efd3e18) rather than the hub id, so pass the base model explicitly as above. The Gemma adapters were trained on the multimodal checkpoints' language model (class Gemma3ForConditionalGeneration); if
|
| 118 |
+
loading through AutoModelForCausalLM leaves modules unmatched, load the base with Gemma3ForConditionalGeneration instead.
|
| 119 |
+
|
| 120 |
+
## License
|
| 121 |
+
|
| 122 |
+
The Gemma adapters are derivatives of Gemma 3 and fall under the Gemma Terms of Use. The Qwen adapters are derivatives of Qwen3-14B
|
| 123 |
+
(Apache 2.0).
|
gemma-3-27b-it/bottom-s0/README.md
ADDED
|
@@ -0,0 +1,206 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: google/gemma-3-27b-it
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- base_model:adapter:google/gemma-3-27b-it
|
| 6 |
+
- lora
|
| 7 |
+
- transformers
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# Model Card for Model ID
|
| 11 |
+
|
| 12 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
## Model Details
|
| 17 |
+
|
| 18 |
+
### Model Description
|
| 19 |
+
|
| 20 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
- **Developed by:** [More Information Needed]
|
| 25 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 26 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 27 |
+
- **Model type:** [More Information Needed]
|
| 28 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 29 |
+
- **License:** [More Information Needed]
|
| 30 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 31 |
+
|
| 32 |
+
### Model Sources [optional]
|
| 33 |
+
|
| 34 |
+
<!-- Provide the basic links for the model. -->
|
| 35 |
+
|
| 36 |
+
- **Repository:** [More Information Needed]
|
| 37 |
+
- **Paper [optional]:** [More Information Needed]
|
| 38 |
+
- **Demo [optional]:** [More Information Needed]
|
| 39 |
+
|
| 40 |
+
## Uses
|
| 41 |
+
|
| 42 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 43 |
+
|
| 44 |
+
### Direct Use
|
| 45 |
+
|
| 46 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 47 |
+
|
| 48 |
+
[More Information Needed]
|
| 49 |
+
|
| 50 |
+
### Downstream Use [optional]
|
| 51 |
+
|
| 52 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 53 |
+
|
| 54 |
+
[More Information Needed]
|
| 55 |
+
|
| 56 |
+
### Out-of-Scope Use
|
| 57 |
+
|
| 58 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 59 |
+
|
| 60 |
+
[More Information Needed]
|
| 61 |
+
|
| 62 |
+
## Bias, Risks, and Limitations
|
| 63 |
+
|
| 64 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 65 |
+
|
| 66 |
+
[More Information Needed]
|
| 67 |
+
|
| 68 |
+
### Recommendations
|
| 69 |
+
|
| 70 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 71 |
+
|
| 72 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 73 |
+
|
| 74 |
+
## How to Get Started with the Model
|
| 75 |
+
|
| 76 |
+
Use the code below to get started with the model.
|
| 77 |
+
|
| 78 |
+
[More Information Needed]
|
| 79 |
+
|
| 80 |
+
## Training Details
|
| 81 |
+
|
| 82 |
+
### Training Data
|
| 83 |
+
|
| 84 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 85 |
+
|
| 86 |
+
[More Information Needed]
|
| 87 |
+
|
| 88 |
+
### Training Procedure
|
| 89 |
+
|
| 90 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 91 |
+
|
| 92 |
+
#### Preprocessing [optional]
|
| 93 |
+
|
| 94 |
+
[More Information Needed]
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
#### Training Hyperparameters
|
| 98 |
+
|
| 99 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 100 |
+
|
| 101 |
+
#### Speeds, Sizes, Times [optional]
|
| 102 |
+
|
| 103 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 104 |
+
|
| 105 |
+
[More Information Needed]
|
| 106 |
+
|
| 107 |
+
## Evaluation
|
| 108 |
+
|
| 109 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 110 |
+
|
| 111 |
+
### Testing Data, Factors & Metrics
|
| 112 |
+
|
| 113 |
+
#### Testing Data
|
| 114 |
+
|
| 115 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 116 |
+
|
| 117 |
+
[More Information Needed]
|
| 118 |
+
|
| 119 |
+
#### Factors
|
| 120 |
+
|
| 121 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 122 |
+
|
| 123 |
+
[More Information Needed]
|
| 124 |
+
|
| 125 |
+
#### Metrics
|
| 126 |
+
|
| 127 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 128 |
+
|
| 129 |
+
[More Information Needed]
|
| 130 |
+
|
| 131 |
+
### Results
|
| 132 |
+
|
| 133 |
+
[More Information Needed]
|
| 134 |
+
|
| 135 |
+
#### Summary
|
| 136 |
+
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
## Model Examination [optional]
|
| 140 |
+
|
| 141 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 142 |
+
|
| 143 |
+
[More Information Needed]
|
| 144 |
+
|
| 145 |
+
## Environmental Impact
|
| 146 |
+
|
| 147 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 148 |
+
|
| 149 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 150 |
+
|
| 151 |
+
- **Hardware Type:** [More Information Needed]
|
| 152 |
+
- **Hours used:** [More Information Needed]
|
| 153 |
+
- **Cloud Provider:** [More Information Needed]
|
| 154 |
+
- **Compute Region:** [More Information Needed]
|
| 155 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 156 |
+
|
| 157 |
+
## Technical Specifications [optional]
|
| 158 |
+
|
| 159 |
+
### Model Architecture and Objective
|
| 160 |
+
|
| 161 |
+
[More Information Needed]
|
| 162 |
+
|
| 163 |
+
### Compute Infrastructure
|
| 164 |
+
|
| 165 |
+
[More Information Needed]
|
| 166 |
+
|
| 167 |
+
#### Hardware
|
| 168 |
+
|
| 169 |
+
[More Information Needed]
|
| 170 |
+
|
| 171 |
+
#### Software
|
| 172 |
+
|
| 173 |
+
[More Information Needed]
|
| 174 |
+
|
| 175 |
+
## Citation [optional]
|
| 176 |
+
|
| 177 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 178 |
+
|
| 179 |
+
**BibTeX:**
|
| 180 |
+
|
| 181 |
+
[More Information Needed]
|
| 182 |
+
|
| 183 |
+
**APA:**
|
| 184 |
+
|
| 185 |
+
[More Information Needed]
|
| 186 |
+
|
| 187 |
+
## Glossary [optional]
|
| 188 |
+
|
| 189 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 190 |
+
|
| 191 |
+
[More Information Needed]
|
| 192 |
+
|
| 193 |
+
## More Information [optional]
|
| 194 |
+
|
| 195 |
+
[More Information Needed]
|
| 196 |
+
|
| 197 |
+
## Model Card Authors [optional]
|
| 198 |
+
|
| 199 |
+
[More Information Needed]
|
| 200 |
+
|
| 201 |
+
## Model Card Contact
|
| 202 |
+
|
| 203 |
+
[More Information Needed]
|
| 204 |
+
### Framework versions
|
| 205 |
+
|
| 206 |
+
- PEFT 0.17.1
|
gemma-3-27b-it/bottom-s0/adapter_config.json
ADDED
|
@@ -0,0 +1,228 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": {
|
| 4 |
+
"base_model_class": "Gemma3ForConditionalGeneration",
|
| 5 |
+
"parent_library": "transformers.models.gemma3.modeling_gemma3"
|
| 6 |
+
},
|
| 7 |
+
"base_model_name_or_path": "google/gemma-3-27b-it",
|
| 8 |
+
"bias": "none",
|
| 9 |
+
"corda_config": null,
|
| 10 |
+
"eva_config": null,
|
| 11 |
+
"exclude_modules": null,
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": true,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layer_replication": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"layers_to_transform": null,
|
| 18 |
+
"loftq_config": {},
|
| 19 |
+
"lora_alpha": 128,
|
| 20 |
+
"lora_bias": false,
|
| 21 |
+
"lora_dropout": 0.05,
|
| 22 |
+
"megatron_config": null,
|
| 23 |
+
"megatron_core": "megatron.core",
|
| 24 |
+
"modules_to_save": null,
|
| 25 |
+
"peft_type": "LORA",
|
| 26 |
+
"qalora_group_size": 16,
|
| 27 |
+
"r": 64,
|
| 28 |
+
"rank_pattern": {},
|
| 29 |
+
"revision": null,
|
| 30 |
+
"target_modules": [
|
| 31 |
+
"language_model.layers.13.self_attn.v_proj",
|
| 32 |
+
"language_model.layers.14.self_attn.q_proj",
|
| 33 |
+
"language_model.layers.26.self_attn.k_proj",
|
| 34 |
+
"53.self_attn.q_proj",
|
| 35 |
+
"42.self_attn.v_proj",
|
| 36 |
+
"27.self_attn.q_proj",
|
| 37 |
+
"language_model.layers.18.self_attn.v_proj",
|
| 38 |
+
"28.self_attn.k_proj",
|
| 39 |
+
"57.self_attn.q_proj",
|
| 40 |
+
"language_model.layers.11.self_attn.q_proj",
|
| 41 |
+
"29.self_attn.k_proj",
|
| 42 |
+
"31.self_attn.q_proj",
|
| 43 |
+
"58.self_attn.k_proj",
|
| 44 |
+
"language_model.layers.1.self_attn.k_proj",
|
| 45 |
+
"language_model.layers.11.self_attn.v_proj",
|
| 46 |
+
"language_model.layers.17.self_attn.q_proj",
|
| 47 |
+
"46.self_attn.q_proj",
|
| 48 |
+
"language_model.layers.7.self_attn.v_proj",
|
| 49 |
+
"35.self_attn.v_proj",
|
| 50 |
+
"45.self_attn.k_proj",
|
| 51 |
+
"language_model.layers.3.self_attn.v_proj",
|
| 52 |
+
"27.self_attn.v_proj",
|
| 53 |
+
"53.self_attn.v_proj",
|
| 54 |
+
"language_model.layers.5.self_attn.q_proj",
|
| 55 |
+
"o_proj",
|
| 56 |
+
"27.self_attn.k_proj",
|
| 57 |
+
"33.self_attn.q_proj",
|
| 58 |
+
"36.self_attn.q_proj",
|
| 59 |
+
"49.self_attn.q_proj",
|
| 60 |
+
"language_model.layers.22.self_attn.v_proj",
|
| 61 |
+
"language_model.layers.25.self_attn.q_proj",
|
| 62 |
+
"language_model.layers.13.self_attn.q_proj",
|
| 63 |
+
"language_model.layers.7.self_attn.q_proj",
|
| 64 |
+
"language_model.layers.4.self_attn.q_proj",
|
| 65 |
+
"58.self_attn.q_proj",
|
| 66 |
+
"28.self_attn.q_proj",
|
| 67 |
+
"language_model.layers.10.self_attn.q_proj",
|
| 68 |
+
"61.self_attn.k_proj",
|
| 69 |
+
"34.self_attn.v_proj",
|
| 70 |
+
"56.self_attn.v_proj",
|
| 71 |
+
"36.self_attn.k_proj",
|
| 72 |
+
"32.self_attn.k_proj",
|
| 73 |
+
"32.self_attn.q_proj",
|
| 74 |
+
"30.self_attn.k_proj",
|
| 75 |
+
"language_model.layers.16.self_attn.q_proj",
|
| 76 |
+
"language_model.layers.0.self_attn.q_proj",
|
| 77 |
+
"language_model.layers.1.self_attn.q_proj",
|
| 78 |
+
"language_model.layers.22.self_attn.q_proj",
|
| 79 |
+
"language_model.layers.9.self_attn.q_proj",
|
| 80 |
+
"47.self_attn.q_proj",
|
| 81 |
+
"40.self_attn.k_proj",
|
| 82 |
+
"language_model.layers.11.self_attn.k_proj",
|
| 83 |
+
"language_model.layers.15.self_attn.q_proj",
|
| 84 |
+
"language_model.layers.25.self_attn.v_proj",
|
| 85 |
+
"language_model.layers.3.self_attn.k_proj",
|
| 86 |
+
"52.self_attn.k_proj",
|
| 87 |
+
"33.self_attn.k_proj",
|
| 88 |
+
"language_model.layers.18.self_attn.q_proj",
|
| 89 |
+
"50.self_attn.q_proj",
|
| 90 |
+
"35.self_attn.k_proj",
|
| 91 |
+
"language_model.layers.0.self_attn.k_proj",
|
| 92 |
+
"31.self_attn.k_proj",
|
| 93 |
+
"36.self_attn.v_proj",
|
| 94 |
+
"46.self_attn.v_proj",
|
| 95 |
+
"language_model.layers.6.self_attn.q_proj",
|
| 96 |
+
"38.self_attn.v_proj",
|
| 97 |
+
"language_model.layers.19.self_attn.q_proj",
|
| 98 |
+
"60.self_attn.q_proj",
|
| 99 |
+
"38.self_attn.q_proj",
|
| 100 |
+
"54.self_attn.q_proj",
|
| 101 |
+
"34.self_attn.k_proj",
|
| 102 |
+
"57.self_attn.v_proj",
|
| 103 |
+
"language_model.layers.26.self_attn.v_proj",
|
| 104 |
+
"language_model.layers.23.self_attn.k_proj",
|
| 105 |
+
"language_model.layers.21.self_attn.q_proj",
|
| 106 |
+
"47.self_attn.v_proj",
|
| 107 |
+
"55.self_attn.v_proj",
|
| 108 |
+
"51.self_attn.q_proj",
|
| 109 |
+
"37.self_attn.k_proj",
|
| 110 |
+
"42.self_attn.q_proj",
|
| 111 |
+
"language_model.layers.13.self_attn.k_proj",
|
| 112 |
+
"language_model.layers.16.self_attn.v_proj",
|
| 113 |
+
"language_model.layers.8.self_attn.k_proj",
|
| 114 |
+
"40.self_attn.q_proj",
|
| 115 |
+
"language_model.layers.25.self_attn.k_proj",
|
| 116 |
+
"language_model.layers.26.self_attn.q_proj",
|
| 117 |
+
"37.self_attn.v_proj",
|
| 118 |
+
"28.self_attn.v_proj",
|
| 119 |
+
"language_model.layers.17.self_attn.v_proj",
|
| 120 |
+
"52.self_attn.v_proj",
|
| 121 |
+
"language_model.layers.24.self_attn.v_proj",
|
| 122 |
+
"39.self_attn.q_proj",
|
| 123 |
+
"29.self_attn.v_proj",
|
| 124 |
+
"language_model.layers.16.self_attn.k_proj",
|
| 125 |
+
"40.self_attn.v_proj",
|
| 126 |
+
"41.self_attn.k_proj",
|
| 127 |
+
"48.self_attn.q_proj",
|
| 128 |
+
"37.self_attn.q_proj",
|
| 129 |
+
"language_model.layers.3.self_attn.q_proj",
|
| 130 |
+
"48.self_attn.k_proj",
|
| 131 |
+
"57.self_attn.k_proj",
|
| 132 |
+
"34.self_attn.q_proj",
|
| 133 |
+
"56.self_attn.k_proj",
|
| 134 |
+
"43.self_attn.k_proj",
|
| 135 |
+
"language_model.layers.23.self_attn.q_proj",
|
| 136 |
+
"language_model.layers.9.self_attn.v_proj",
|
| 137 |
+
"language_model.layers.8.self_attn.v_proj",
|
| 138 |
+
"30.self_attn.q_proj",
|
| 139 |
+
"60.self_attn.k_proj",
|
| 140 |
+
"language_model.layers.9.self_attn.k_proj",
|
| 141 |
+
"39.self_attn.v_proj",
|
| 142 |
+
"61.self_attn.v_proj",
|
| 143 |
+
"language_model.layers.7.self_attn.k_proj",
|
| 144 |
+
"59.self_attn.v_proj",
|
| 145 |
+
"language_model.layers.4.self_attn.k_proj",
|
| 146 |
+
"42.self_attn.k_proj",
|
| 147 |
+
"language_model.layers.20.self_attn.q_proj",
|
| 148 |
+
"55.self_attn.q_proj",
|
| 149 |
+
"55.self_attn.k_proj",
|
| 150 |
+
"38.self_attn.k_proj",
|
| 151 |
+
"down_proj",
|
| 152 |
+
"60.self_attn.v_proj",
|
| 153 |
+
"52.self_attn.q_proj",
|
| 154 |
+
"gate_proj",
|
| 155 |
+
"language_model.layers.6.self_attn.k_proj",
|
| 156 |
+
"language_model.layers.0.self_attn.v_proj",
|
| 157 |
+
"61.self_attn.q_proj",
|
| 158 |
+
"29.self_attn.q_proj",
|
| 159 |
+
"51.self_attn.k_proj",
|
| 160 |
+
"up_proj",
|
| 161 |
+
"31.self_attn.v_proj",
|
| 162 |
+
"language_model.layers.12.self_attn.k_proj",
|
| 163 |
+
"language_model.layers.21.self_attn.v_proj",
|
| 164 |
+
"language_model.layers.20.self_attn.k_proj",
|
| 165 |
+
"language_model.layers.10.self_attn.k_proj",
|
| 166 |
+
"43.self_attn.v_proj",
|
| 167 |
+
"48.self_attn.v_proj",
|
| 168 |
+
"49.self_attn.v_proj",
|
| 169 |
+
"language_model.layers.14.self_attn.v_proj",
|
| 170 |
+
"language_model.layers.14.self_attn.k_proj",
|
| 171 |
+
"language_model.layers.5.self_attn.v_proj",
|
| 172 |
+
"43.self_attn.q_proj",
|
| 173 |
+
"58.self_attn.v_proj",
|
| 174 |
+
"language_model.layers.12.self_attn.v_proj",
|
| 175 |
+
"language_model.layers.2.self_attn.q_proj",
|
| 176 |
+
"language_model.layers.24.self_attn.k_proj",
|
| 177 |
+
"47.self_attn.k_proj",
|
| 178 |
+
"language_model.layers.5.self_attn.k_proj",
|
| 179 |
+
"33.self_attn.v_proj",
|
| 180 |
+
"language_model.layers.18.self_attn.k_proj",
|
| 181 |
+
"language_model.layers.15.self_attn.k_proj",
|
| 182 |
+
"language_model.layers.2.self_attn.v_proj",
|
| 183 |
+
"50.self_attn.k_proj",
|
| 184 |
+
"language_model.layers.19.self_attn.v_proj",
|
| 185 |
+
"49.self_attn.k_proj",
|
| 186 |
+
"language_model.layers.24.self_attn.q_proj",
|
| 187 |
+
"language_model.layers.21.self_attn.k_proj",
|
| 188 |
+
"language_model.layers.23.self_attn.v_proj",
|
| 189 |
+
"language_model.layers.1.self_attn.v_proj",
|
| 190 |
+
"language_model.layers.10.self_attn.v_proj",
|
| 191 |
+
"language_model.layers.20.self_attn.v_proj",
|
| 192 |
+
"45.self_attn.v_proj",
|
| 193 |
+
"30.self_attn.v_proj",
|
| 194 |
+
"language_model.layers.6.self_attn.v_proj",
|
| 195 |
+
"45.self_attn.q_proj",
|
| 196 |
+
"language_model.layers.12.self_attn.q_proj",
|
| 197 |
+
"59.self_attn.k_proj",
|
| 198 |
+
"41.self_attn.q_proj",
|
| 199 |
+
"language_model.layers.17.self_attn.k_proj",
|
| 200 |
+
"41.self_attn.v_proj",
|
| 201 |
+
"39.self_attn.k_proj",
|
| 202 |
+
"44.self_attn.v_proj",
|
| 203 |
+
"44.self_attn.k_proj",
|
| 204 |
+
"language_model.layers.8.self_attn.q_proj",
|
| 205 |
+
"50.self_attn.v_proj",
|
| 206 |
+
"language_model.layers.2.self_attn.k_proj",
|
| 207 |
+
"56.self_attn.q_proj",
|
| 208 |
+
"language_model.layers.15.self_attn.v_proj",
|
| 209 |
+
"51.self_attn.v_proj",
|
| 210 |
+
"32.self_attn.v_proj",
|
| 211 |
+
"53.self_attn.k_proj",
|
| 212 |
+
"54.self_attn.k_proj",
|
| 213 |
+
"35.self_attn.q_proj",
|
| 214 |
+
"language_model.layers.22.self_attn.k_proj",
|
| 215 |
+
"54.self_attn.v_proj",
|
| 216 |
+
"44.self_attn.q_proj",
|
| 217 |
+
"language_model.layers.4.self_attn.v_proj",
|
| 218 |
+
"language_model.layers.19.self_attn.k_proj",
|
| 219 |
+
"46.self_attn.k_proj",
|
| 220 |
+
"59.self_attn.q_proj"
|
| 221 |
+
],
|
| 222 |
+
"target_parameters": null,
|
| 223 |
+
"task_type": null,
|
| 224 |
+
"trainable_token_indices": null,
|
| 225 |
+
"use_dora": false,
|
| 226 |
+
"use_qalora": false,
|
| 227 |
+
"use_rslora": false
|
| 228 |
+
}
|
gemma-3-27b-it/bottom-s0/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bd41c484fcca91d0d333f3d17cc3e749b0c69730171f174515e16a0eb4b3ad60
|
| 3 |
+
size 1816396048
|
gemma-3-27b-it/bottom-s0/config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "google/gemma-3-27b-it",
|
| 3 |
+
"data": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma/arms/TASKW-bottom-gemma.jsonl",
|
| 4 |
+
"run": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma/runs/bottom-s0",
|
| 5 |
+
"steps": 0,
|
| 6 |
+
"batch": 16,
|
| 7 |
+
"micro": 1,
|
| 8 |
+
"seq_len": 4096,
|
| 9 |
+
"lr": 0.0001,
|
| 10 |
+
"warmup": 10,
|
| 11 |
+
"lora_rank": 64,
|
| 12 |
+
"lora_targets": "all-linear",
|
| 13 |
+
"lora_targets_file": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/code/gemma3_lora_targets.json",
|
| 14 |
+
"seed": 0,
|
| 15 |
+
"bos": true,
|
| 16 |
+
"no_merge": false,
|
| 17 |
+
"save_at": "",
|
| 18 |
+
"replay": "",
|
| 19 |
+
"replay_frac": 0.0,
|
| 20 |
+
"constant_lr": false,
|
| 21 |
+
"schedule": "hf-cosine",
|
| 22 |
+
"min_lr_frac": 0.0,
|
| 23 |
+
"warmup_ratio": 0.05,
|
| 24 |
+
"lora_dropout": 0.05,
|
| 25 |
+
"no_pack": true,
|
| 26 |
+
"chat_sft": true,
|
| 27 |
+
"data_frac": 1.0,
|
| 28 |
+
"base_adapter": "",
|
| 29 |
+
"ckpt_every": 25,
|
| 30 |
+
"kill_after": 0,
|
| 31 |
+
"method": "doc_lm_lora",
|
| 32 |
+
"world": 1
|
| 33 |
+
}
|
gemma-3-27b-it/bottom-s0/log.jsonl
ADDED
|
@@ -0,0 +1,143 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 1, "loss": 0.2388627216532223, "grad_norm": 0.21762774884700775, "lr": 2.857142857142857e-05, "epoch": 1, "t": 12.444128513336182}
|
| 2 |
+
{"step": 2, "loss": 0.235712662135085, "grad_norm": 0.2515455484390259, "lr": 4.2857142857142856e-05, "epoch": 1, "t": 11.325777292251587}
|
| 3 |
+
{"step": 3, "loss": 0.20722852311133905, "grad_norm": 0.24037572741508484, "lr": 5.714285714285714e-05, "epoch": 1, "t": 9.33599042892456}
|
| 4 |
+
{"step": 4, "loss": 0.2143341991949213, "grad_norm": 0.23092089593410492, "lr": 7.142857142857143e-05, "epoch": 1, "t": 11.79221248626709}
|
| 5 |
+
{"step": 5, "loss": 0.23137090081444114, "grad_norm": 0.28006842732429504, "lr": 8.571428571428571e-05, "epoch": 1, "t": 10.478177070617676}
|
| 6 |
+
{"step": 6, "loss": 0.20194122198205314, "grad_norm": 0.31564444303512573, "lr": 0.0001, "epoch": 1, "t": 11.174200057983398}
|
| 7 |
+
{"step": 7, "loss": 0.2239803262572736, "grad_norm": 0.26979243755340576, "lr": 0.0001, "epoch": 1, "t": 10.888554573059082}
|
| 8 |
+
{"step": 8, "loss": 0.23328136752533052, "grad_norm": 0.2810402810573578, "lr": 9.998666040558187e-05, "epoch": 1, "t": 9.733090162277222}
|
| 9 |
+
{"step": 9, "loss": 0.22849376963830859, "grad_norm": 0.27494359016418457, "lr": 9.994664874011863e-05, "epoch": 1, "t": 9.6326162815094}
|
| 10 |
+
{"step": 10, "loss": 0.21287972637334324, "grad_norm": 0.2550443112850189, "lr": 9.987998635318585e-05, "epoch": 1, "t": 12.759542226791382}
|
| 11 |
+
{"step": 11, "loss": 0.1982550655120606, "grad_norm": 0.23410779237747192, "lr": 9.978670881475172e-05, "epoch": 1, "t": 9.869225025177002}
|
| 12 |
+
{"step": 12, "loss": 0.21076927560160313, "grad_norm": 0.2532612681388855, "lr": 9.966686589619751e-05, "epoch": 1, "t": 10.686810970306396}
|
| 13 |
+
{"step": 13, "loss": 0.22913969583991511, "grad_norm": 0.23565331101417542, "lr": 9.952052154376026e-05, "epoch": 1, "t": 10.424184083938599}
|
| 14 |
+
{"step": 14, "loss": 0.22744822150029614, "grad_norm": 0.2481132298707962, "lr": 9.934775384441228e-05, "epoch": 1, "t": 10.61027479171753}
|
| 15 |
+
{"step": 15, "loss": 0.22230122700802998, "grad_norm": 0.24708789587020874, "lr": 9.91486549841951e-05, "epoch": 1, "t": 10.84962248802185}
|
| 16 |
+
{"step": 16, "loss": 0.21122691495941942, "grad_norm": 0.26503968238830566, "lr": 9.892333119903046e-05, "epoch": 1, "t": 8.985432624816895}
|
| 17 |
+
{"step": 17, "loss": 0.2224858362656053, "grad_norm": 0.3172319531440735, "lr": 9.867190271803465e-05, "epoch": 1, "t": 10.85428500175476}
|
| 18 |
+
{"step": 18, "loss": 0.22527228779039882, "grad_norm": 0.2662495970726013, "lr": 9.839450369936613e-05, "epoch": 1, "t": 9.51560640335083}
|
| 19 |
+
{"step": 19, "loss": 0.21141741523108248, "grad_norm": 0.23125584423542023, "lr": 9.809128215864097e-05, "epoch": 1, "t": 12.469651937484741}
|
| 20 |
+
{"step": 20, "loss": 0.20232847735778844, "grad_norm": 0.23719441890716553, "lr": 9.7762399889954e-05, "epoch": 1, "t": 11.577425479888916}
|
| 21 |
+
{"step": 21, "loss": 0.20107995960074412, "grad_norm": 0.24122916162014008, "lr": 9.74080323795483e-05, "epoch": 1, "t": 10.272185802459717}
|
| 22 |
+
{"step": 22, "loss": 0.17664560980559602, "grad_norm": 0.25398024916648865, "lr": 9.702836871217839e-05, "epoch": 1, "t": 10.448529481887817}
|
| 23 |
+
{"step": 23, "loss": 0.19546685274029382, "grad_norm": 0.2274629920721054, "lr": 9.662361147021779e-05, "epoch": 1, "t": 10.469770908355713}
|
| 24 |
+
{"step": 24, "loss": 0.21242517471666392, "grad_norm": 0.2642868459224701, "lr": 9.619397662556435e-05, "epoch": 1, "t": 9.797579288482666}
|
| 25 |
+
{"step": 25, "loss": 0.20459047934479205, "grad_norm": 0.2398044615983963, "lr": 9.573969342440106e-05, "epoch": 1, "t": 9.705522537231445}
|
| 26 |
+
{"step": 26, "loss": 0.23657778518899333, "grad_norm": 0.2524448037147522, "lr": 9.526100426487409e-05, "epoch": 1, "t": 11.261531352996826}
|
| 27 |
+
{"step": 27, "loss": 0.19602776081321682, "grad_norm": 0.20272652804851532, "lr": 9.475816456775313e-05, "epoch": 1, "t": 10.875118732452393}
|
| 28 |
+
{"step": 28, "loss": 0.2211669640385928, "grad_norm": 0.19873002171516418, "lr": 9.423144264014277e-05, "epoch": 1, "t": 11.876128435134888}
|
| 29 |
+
{"step": 29, "loss": 0.1947025089577412, "grad_norm": 0.17811954021453857, "lr": 9.368111953231848e-05, "epoch": 1, "t": 11.735994338989258}
|
| 30 |
+
{"step": 30, "loss": 0.21522573809763904, "grad_norm": 0.22015735507011414, "lr": 9.310748888776254e-05, "epoch": 1, "t": 9.32709550857544}
|
| 31 |
+
{"step": 31, "loss": 0.20544734347410715, "grad_norm": 0.27868083119392395, "lr": 9.251085678648072e-05, "epoch": 1, "t": 8.893455266952515}
|
| 32 |
+
{"step": 32, "loss": 0.23583441675494377, "grad_norm": 0.2901236116886139, "lr": 9.189154158168292e-05, "epoch": 1, "t": 9.24516749382019}
|
| 33 |
+
{"step": 33, "loss": 0.21743471025751626, "grad_norm": 0.22271491587162018, "lr": 9.124987372991511e-05, "epoch": 1, "t": 11.855639219284058}
|
| 34 |
+
{"step": 34, "loss": 0.23717964832399452, "grad_norm": 0.21913611888885498, "lr": 9.058619561473307e-05, "epoch": 1, "t": 11.726173400878906}
|
| 35 |
+
{"step": 35, "loss": 0.20759814868413645, "grad_norm": 0.23411060869693756, "lr": 8.9900861364012e-05, "epoch": 1, "t": 10.314544200897217}
|
| 36 |
+
{"step": 36, "loss": 0.23809238732337423, "grad_norm": 0.2573903203010559, "lr": 8.919423666098969e-05, "epoch": 1, "t": 11.00603199005127}
|
| 37 |
+
{"step": 37, "loss": 0.24915506220712397, "grad_norm": 0.21866685152053833, "lr": 8.846669854914396e-05, "epoch": 1, "t": 12.01134991645813}
|
| 38 |
+
{"step": 38, "loss": 0.21759839139853146, "grad_norm": 0.22691188752651215, "lr": 8.77186352310082e-05, "epoch": 1, "t": 10.31660771369934}
|
| 39 |
+
{"step": 39, "loss": 0.2030508272627835, "grad_norm": 0.26183345913887024, "lr": 8.695044586103296e-05, "epoch": 1, "t": 8.541102409362793}
|
| 40 |
+
{"step": 40, "loss": 0.22065777100652237, "grad_norm": 0.21961724758148193, "lr": 8.616254033260351e-05, "epoch": 1, "t": 9.856967687606812}
|
| 41 |
+
{"step": 41, "loss": 0.21872945046281175, "grad_norm": 0.24082328379154205, "lr": 8.535533905932738e-05, "epoch": 1, "t": 9.37130093574524}
|
| 42 |
+
{"step": 42, "loss": 0.20822276412936347, "grad_norm": 0.2052266150712967, "lr": 8.452927275070858e-05, "epoch": 1, "t": 11.595879793167114}
|
| 43 |
+
{"step": 43, "loss": 0.19798607559768294, "grad_norm": 0.1957789957523346, "lr": 8.368478218232787e-05, "epoch": 1, "t": 10.872885465621948}
|
| 44 |
+
{"step": 44, "loss": 0.22892043988692107, "grad_norm": 0.19867897033691406, "lr": 8.282231796065214e-05, "epoch": 1, "t": 10.244351387023926}
|
| 45 |
+
{"step": 45, "loss": 0.242736388089364, "grad_norm": 0.27122998237609863, "lr": 8.194234028259806e-05, "epoch": 1, "t": 8.18606686592102}
|
| 46 |
+
{"step": 46, "loss": 0.21852992940529564, "grad_norm": 0.2174919992685318, "lr": 8.104531868997858e-05, "epoch": 1, "t": 10.524156332015991}
|
| 47 |
+
{"step": 47, "loss": 0.22686265551998153, "grad_norm": 0.25648730993270874, "lr": 8.013173181896283e-05, "epoch": 1, "t": 9.041518926620483}
|
| 48 |
+
{"step": 48, "loss": 0.2029904941701377, "grad_norm": 0.22947531938552856, "lr": 7.920206714468383e-05, "epoch": 1, "t": 10.2853844165802}
|
| 49 |
+
{"step": 49, "loss": 0.19985105391312444, "grad_norm": 0.22045272588729858, "lr": 7.82568207211296e-05, "epoch": 1, "t": 11.130200862884521}
|
| 50 |
+
{"step": 50, "loss": 0.22209859944843813, "grad_norm": 0.26223480701446533, "lr": 7.729649691645673e-05, "epoch": 1, "t": 10.229337453842163}
|
| 51 |
+
{"step": 51, "loss": 0.20452746713266892, "grad_norm": 0.21888244152069092, "lr": 7.63216081438678e-05, "epoch": 1, "t": 9.349253416061401}
|
| 52 |
+
{"step": 52, "loss": 0.22174791616875833, "grad_norm": 0.21392081677913666, "lr": 7.533267458819598e-05, "epoch": 1, "t": 11.198519945144653}
|
| 53 |
+
{"step": 53, "loss": 0.21814539286044324, "grad_norm": 0.22639450430870056, "lr": 7.433022392834282e-05, "epoch": 1, "t": 10.61546516418457}
|
| 54 |
+
{"step": 54, "loss": 0.24107612475326792, "grad_norm": 0.26464715600013733, "lr": 7.33147910557174e-05, "epoch": 1, "t": 10.002691745758057}
|
| 55 |
+
{"step": 55, "loss": 0.22465880092330429, "grad_norm": 0.23496244847774506, "lr": 7.228691778882693e-05, "epoch": 1, "t": 10.007191896438599}
|
| 56 |
+
{"step": 56, "loss": 0.22600633462993341, "grad_norm": 0.24210432171821594, "lr": 7.12471525841711e-05, "epoch": 1, "t": 9.720618724822998}
|
| 57 |
+
{"step": 57, "loss": 0.21559080427409902, "grad_norm": 0.23423489928245544, "lr": 7.019605024359474e-05, "epoch": 1, "t": 10.691621541976929}
|
| 58 |
+
{"step": 58, "loss": 0.19947587747568638, "grad_norm": 0.20437510311603546, "lr": 6.91341716182545e-05, "epoch": 1, "t": 10.225821256637573}
|
| 59 |
+
{"step": 59, "loss": 0.2269615984508082, "grad_norm": 0.204420804977417, "lr": 6.806208330935766e-05, "epoch": 1, "t": 11.657158851623535}
|
| 60 |
+
{"step": 60, "loss": 0.20841974605132066, "grad_norm": 0.2264527827501297, "lr": 6.698035736583306e-05, "epoch": 1, "t": 11.331000089645386}
|
| 61 |
+
{"step": 61, "loss": 0.20616325071715255, "grad_norm": 0.2029811292886734, "lr": 6.588957097909508e-05, "epoch": 1, "t": 9.483571767807007}
|
| 62 |
+
{"step": 62, "loss": 0.2324178780736628, "grad_norm": 0.24112142622470856, "lr": 6.479030617506353e-05, "epoch": 1, "t": 8.882867574691772}
|
| 63 |
+
{"step": 63, "loss": 0.23443861779852496, "grad_norm": 0.22643259167671204, "lr": 6.368314950360415e-05, "epoch": 1, "t": 9.968997955322266}
|
| 64 |
+
{"step": 64, "loss": 0.22357680476137062, "grad_norm": 0.2048519402742386, "lr": 6.256869172555515e-05, "epoch": 1, "t": 9.908613443374634}
|
| 65 |
+
{"step": 65, "loss": 0.22116600295120403, "grad_norm": 0.2066519558429718, "lr": 6.14475274975067e-05, "epoch": 1, "t": 12.06965184211731}
|
| 66 |
+
{"step": 66, "loss": 0.21236905787615762, "grad_norm": 0.22835209965705872, "lr": 6.0320255054501985e-05, "epoch": 1, "t": 10.543720483779907}
|
| 67 |
+
{"step": 67, "loss": 0.2371175207897513, "grad_norm": 0.247774139046669, "lr": 5.918747589082853e-05, "epoch": 1, "t": 10.864552021026611}
|
| 68 |
+
{"step": 68, "loss": 0.2197630648005709, "grad_norm": 0.227577805519104, "lr": 5.8049794439070646e-05, "epoch": 1, "t": 10.829199075698853}
|
| 69 |
+
{"step": 69, "loss": 0.1880163135430921, "grad_norm": 0.1889297366142273, "lr": 5.6907817747594116e-05, "epoch": 1, "t": 12.006531238555908}
|
| 70 |
+
{"step": 70, "loss": 0.23173571132531132, "grad_norm": 0.22342267632484436, "lr": 5.576215515663489e-05, "epoch": 1, "t": 10.81044340133667}
|
| 71 |
+
{"step": 71, "loss": 0.21954204237277283, "grad_norm": 0.19548137485980988, "lr": 5.4613417973165106e-05, "epoch": 1, "t": 11.547603130340576}
|
| 72 |
+
{"step": 72, "loss": 0.20649732471585952, "grad_norm": 0.22420446574687958, "lr": 5.346221914470959e-05, "epoch": 1, "t": 9.054604530334473}
|
| 73 |
+
{"step": 73, "loss": 0.24544292496572623, "grad_norm": 0.2194955199956894, "lr": 5.230917293228699e-05, "epoch": 1, "t": 10.802698135375977}
|
| 74 |
+
{"step": 74, "loss": 0.20944364580356598, "grad_norm": 0.23314836621284485, "lr": 5.115489458265006e-05, "epoch": 1, "t": 12.326721429824829}
|
| 75 |
+
{"step": 75, "loss": 0.20837012654361592, "grad_norm": 0.2116249054670334, "lr": 5e-05, "epoch": 1, "t": 10.97861647605896}
|
| 76 |
+
{"step": 76, "loss": 0.21925544189675483, "grad_norm": 0.24239246547222137, "lr": 4.884510541734995e-05, "epoch": 1, "t": 8.826806783676147}
|
| 77 |
+
{"step": 77, "loss": 0.19496954370531633, "grad_norm": 0.21846184134483337, "lr": 4.7690827067713035e-05, "epoch": 1, "t": 10.840333461761475}
|
| 78 |
+
{"step": 78, "loss": 0.20478971150972078, "grad_norm": 0.19472871720790863, "lr": 4.6537780855290426e-05, "epoch": 1, "t": 11.972460985183716}
|
| 79 |
+
{"step": 79, "loss": 0.22232386492298184, "grad_norm": 0.2340972125530243, "lr": 4.5386582026834906e-05, "epoch": 1, "t": 9.270096063613892}
|
| 80 |
+
{"step": 80, "loss": 0.2253903758615608, "grad_norm": 0.21283625066280365, "lr": 4.423784484336513e-05, "epoch": 1, "t": 11.238869667053223}
|
| 81 |
+
{"step": 81, "loss": 0.2606101710065542, "grad_norm": 0.2696358859539032, "lr": 4.30921822524059e-05, "epoch": 1, "t": 9.446866035461426}
|
| 82 |
+
{"step": 82, "loss": 0.22117945974840114, "grad_norm": 0.23444879055023193, "lr": 4.195020556092935e-05, "epoch": 1, "t": 10.433942317962646}
|
| 83 |
+
{"step": 83, "loss": 0.20961276539138196, "grad_norm": 0.2107972353696823, "lr": 4.0812524109171476e-05, "epoch": 1, "t": 10.173856019973755}
|
| 84 |
+
{"step": 84, "loss": 0.24031980774970083, "grad_norm": 0.21101628243923187, "lr": 3.9679744945498026e-05, "epoch": 1, "t": 11.770877361297607}
|
| 85 |
+
{"step": 85, "loss": 0.19605009515342742, "grad_norm": 0.2318771630525589, "lr": 3.855247250249331e-05, "epoch": 1, "t": 9.469485521316528}
|
| 86 |
+
{"step": 86, "loss": 0.205082954413996, "grad_norm": 0.1792505532503128, "lr": 3.7431308274444865e-05, "epoch": 1, "t": 11.578944683074951}
|
| 87 |
+
{"step": 87, "loss": 0.2294765687058748, "grad_norm": 0.21323147416114807, "lr": 3.631685049639586e-05, "epoch": 1, "t": 11.32551097869873}
|
| 88 |
+
{"step": 88, "loss": 0.2042625909629782, "grad_norm": 0.20144854485988617, "lr": 3.5209693824936484e-05, "epoch": 1, "t": 10.990756511688232}
|
| 89 |
+
{"step": 89, "loss": 0.20178896363717228, "grad_norm": 0.22365014255046844, "lr": 3.411042902090492e-05, "epoch": 1, "t": 9.841756105422974}
|
| 90 |
+
{"step": 90, "loss": 0.21408702990291437, "grad_norm": 0.20460145175457, "lr": 3.301964263416693e-05, "epoch": 1, "t": 10.149108409881592}
|
| 91 |
+
{"step": 91, "loss": 0.22524073024777458, "grad_norm": 0.19301243126392365, "lr": 3.1937916690642356e-05, "epoch": 1, "t": 11.086259841918945}
|
| 92 |
+
{"step": 92, "loss": 0.21679884541138394, "grad_norm": 0.23331990838050842, "lr": 3.086582838174551e-05, "epoch": 1, "t": 9.836403846740723}
|
| 93 |
+
{"step": 93, "loss": 0.19236305064672013, "grad_norm": 0.19747748970985413, "lr": 2.980394975640526e-05, "epoch": 1, "t": 10.784894943237305}
|
| 94 |
+
{"step": 94, "loss": 0.20885155426568414, "grad_norm": 0.21054527163505554, "lr": 2.875284741582892e-05, "epoch": 1, "t": 13.62861943244934}
|
| 95 |
+
{"step": 95, "loss": 0.19573524277606516, "grad_norm": 0.19278784096240997, "lr": 2.771308221117309e-05, "epoch": 1, "t": 9.443552732467651}
|
| 96 |
+
{"step": 96, "loss": 0.18412924991176127, "grad_norm": 0.18215766549110413, "lr": 2.668520894428259e-05, "epoch": 1, "t": 11.46628451347351}
|
| 97 |
+
{"step": 97, "loss": 0.21272699984073873, "grad_norm": 0.22405089437961578, "lr": 2.5669776071657192e-05, "epoch": 1, "t": 10.222633361816406}
|
| 98 |
+
{"step": 98, "loss": 0.20284621577588577, "grad_norm": 0.2214292585849762, "lr": 2.4667325411804036e-05, "epoch": 1, "t": 11.002701044082642}
|
| 99 |
+
{"step": 99, "loss": 0.2037704954699762, "grad_norm": 0.18644998967647552, "lr": 2.3678391856132204e-05, "epoch": 1, "t": 11.841276168823242}
|
| 100 |
+
{"step": 100, "loss": 0.19966401977943807, "grad_norm": 0.18259792029857635, "lr": 2.270350308354329e-05, "epoch": 1, "t": 12.482631206512451}
|
| 101 |
+
{"step": 101, "loss": 0.207604855884218, "grad_norm": 0.21623669564723969, "lr": 2.1743179278870407e-05, "epoch": 1, "t": 11.28463864326477}
|
| 102 |
+
{"step": 102, "loss": 0.1851898413421076, "grad_norm": 0.23804441094398499, "lr": 2.079793285531618e-05, "epoch": 1, "t": 8.047940731048584}
|
| 103 |
+
{"step": 103, "loss": 0.20902058570597437, "grad_norm": 0.21048091351985931, "lr": 1.9868268181037185e-05, "epoch": 1, "t": 9.172752141952515}
|
| 104 |
+
{"step": 104, "loss": 0.21010482448125084, "grad_norm": 0.25184381008148193, "lr": 1.8954681310021433e-05, "epoch": 1, "t": 10.382931470870972}
|
| 105 |
+
{"step": 105, "loss": 0.23947778055840419, "grad_norm": 0.22678008675575256, "lr": 1.8057659717401947e-05, "epoch": 1, "t": 13.313580989837646}
|
| 106 |
+
{"step": 106, "loss": 0.20985241757250658, "grad_norm": 0.195218026638031, "lr": 1.7177682039347876e-05, "epoch": 1, "t": 11.391704320907593}
|
| 107 |
+
{"step": 107, "loss": 0.21111298630295114, "grad_norm": 0.20088234543800354, "lr": 1.631521781767214e-05, "epoch": 1, "t": 10.087232112884521}
|
| 108 |
+
{"step": 108, "loss": 0.2128759815609524, "grad_norm": 0.23645447194576263, "lr": 1.5470727249291423e-05, "epoch": 1, "t": 9.496801137924194}
|
| 109 |
+
{"step": 109, "loss": 0.21148287111933542, "grad_norm": 0.2286359667778015, "lr": 1.4644660940672627e-05, "epoch": 1, "t": 10.831878662109375}
|
| 110 |
+
{"step": 110, "loss": 0.21250255811244328, "grad_norm": 0.21266986429691315, "lr": 1.3837459667396518e-05, "epoch": 1, "t": 10.527068138122559}
|
| 111 |
+
{"step": 111, "loss": 0.1976018484950205, "grad_norm": 0.20844131708145142, "lr": 1.3049554138967051e-05, "epoch": 1, "t": 9.82015323638916}
|
| 112 |
+
{"step": 112, "loss": 0.18631815866600468, "grad_norm": 0.23476539552211761, "lr": 1.2281364768991804e-05, "epoch": 1, "t": 9.4940927028656}
|
| 113 |
+
{"step": 113, "loss": 0.23632987710485218, "grad_norm": 0.26979175209999084, "lr": 1.1533301450856054e-05, "epoch": 1, "t": 8.703019857406616}
|
| 114 |
+
{"step": 114, "loss": 0.19443436921515633, "grad_norm": 0.20499086380004883, "lr": 1.0805763339010327e-05, "epoch": 1, "t": 10.87639045715332}
|
| 115 |
+
{"step": 115, "loss": 0.21916144747761807, "grad_norm": 0.2287338525056839, "lr": 1.0099138635988026e-05, "epoch": 1, "t": 8.22764778137207}
|
| 116 |
+
{"step": 116, "loss": 0.210119454183855, "grad_norm": 0.23289357125759125, "lr": 9.41380438526694e-06, "epoch": 1, "t": 10.153291940689087}
|
| 117 |
+
{"step": 117, "loss": 0.2329445473367826, "grad_norm": 0.23880070447921753, "lr": 8.75012627008489e-06, "epoch": 1, "t": 11.197298288345337}
|
| 118 |
+
{"step": 118, "loss": 0.24593448849733518, "grad_norm": 0.21686413884162903, "lr": 8.10845841831709e-06, "epoch": 1, "t": 11.015234470367432}
|
| 119 |
+
{"step": 119, "loss": 0.19765912835174734, "grad_norm": 0.19902922213077545, "lr": 7.489143213519301e-06, "epoch": 1, "t": 10.7857027053833}
|
| 120 |
+
{"step": 120, "loss": 0.2027934530496965, "grad_norm": 0.1999489665031433, "lr": 6.892511112237471e-06, "epoch": 1, "t": 10.073226690292358}
|
| 121 |
+
{"step": 121, "loss": 0.19633952362598076, "grad_norm": 0.20848709344863892, "lr": 6.318880467681526e-06, "epoch": 1, "t": 10.685622215270996}
|
| 122 |
+
{"step": 122, "loss": 0.22220411396614734, "grad_norm": 0.2322714924812317, "lr": 5.768557359857241e-06, "epoch": 1, "t": 10.177714109420776}
|
| 123 |
+
{"step": 123, "loss": 0.1911583237691857, "grad_norm": 0.2264132797718048, "lr": 5.241835432246889e-06, "epoch": 1, "t": 11.277032852172852}
|
| 124 |
+
{"step": 124, "loss": 0.20983040226728938, "grad_norm": 0.1753399819135666, "lr": 4.7389957351258955e-06, "epoch": 1, "t": 13.683649778366089}
|
| 125 |
+
{"step": 125, "loss": 0.21732753905774996, "grad_norm": 0.21540163457393646, "lr": 4.260306575598949e-06, "epoch": 1, "t": 10.573556661605835}
|
| 126 |
+
{"step": 126, "loss": 0.25159642476597566, "grad_norm": 0.24492903053760529, "lr": 3.8060233744356633e-06, "epoch": 1, "t": 9.432594299316406}
|
| 127 |
+
{"step": 127, "loss": 0.18019520100523737, "grad_norm": 0.20389606058597565, "lr": 3.376388529782215e-06, "epoch": 1, "t": 10.018346071243286}
|
| 128 |
+
{"step": 128, "loss": 0.23019322641362477, "grad_norm": 0.2063959538936615, "lr": 2.971631287821619e-06, "epoch": 1, "t": 11.521349430084229}
|
| 129 |
+
{"step": 129, "loss": 0.2302748969357039, "grad_norm": 0.20593863725662231, "lr": 2.591967620451707e-06, "epoch": 1, "t": 9.945902585983276}
|
| 130 |
+
{"step": 130, "loss": 0.1891610284517595, "grad_norm": 0.20624704658985138, "lr": 2.237600110046001e-06, "epoch": 1, "t": 9.904053688049316}
|
| 131 |
+
{"step": 131, "loss": 0.22144327517606754, "grad_norm": 0.19976010918617249, "lr": 1.908717841359048e-06, "epoch": 1, "t": 11.227613925933838}
|
| 132 |
+
{"step": 132, "loss": 0.20974808988741703, "grad_norm": 0.20737121999263763, "lr": 1.6054963006338742e-06, "epoch": 1, "t": 9.146003723144531}
|
| 133 |
+
{"step": 133, "loss": 0.22936650777750822, "grad_norm": 0.228878453373909, "lr": 1.328097281965357e-06, "epoch": 1, "t": 9.889773607254028}
|
| 134 |
+
{"step": 134, "loss": 0.20860788158952442, "grad_norm": 0.21811050176620483, "lr": 1.0766688009695547e-06, "epoch": 1, "t": 9.530797481536865}
|
| 135 |
+
{"step": 135, "loss": 0.20187840770942764, "grad_norm": 0.23246832191944122, "lr": 8.513450158049108e-07, "epoch": 1, "t": 9.34354829788208}
|
| 136 |
+
{"step": 136, "loss": 0.21056196145508044, "grad_norm": 0.21951112151145935, "lr": 6.522461555877213e-07, "epoch": 1, "t": 9.701274156570435}
|
| 137 |
+
{"step": 137, "loss": 0.21995643253999825, "grad_norm": 0.1930866837501526, "lr": 4.794784562397458e-07, "epoch": 1, "t": 9.245867490768433}
|
| 138 |
+
{"step": 138, "loss": 0.2225570285894442, "grad_norm": 0.2062988579273224, "lr": 3.3313410380250154e-07, "epoch": 1, "t": 10.815188884735107}
|
| 139 |
+
{"step": 139, "loss": 0.229074293990263, "grad_norm": 0.2346707135438919, "lr": 2.1329118524827662e-07, "epoch": 1, "t": 9.128573417663574}
|
| 140 |
+
{"step": 140, "loss": 0.24516224958173582, "grad_norm": 0.23325659334659576, "lr": 1.200136468141544e-07, "epoch": 1, "t": 9.708129405975342}
|
| 141 |
+
{"step": 141, "loss": 0.1919095427009196, "grad_norm": 0.20758429169654846, "lr": 5.3351259881379014e-08, "epoch": 1, "t": 9.917202472686768}
|
| 142 |
+
{"step": 142, "loss": 0.2715901723644077, "grad_norm": 0.251455157995224, "lr": 1.3339594418138035e-08, "epoch": 1, "t": 9.803221702575684}
|
| 143 |
+
{"step": 143, "loss": 0.21295721046610364, "grad_norm": 0.2040252685546875, "lr": 0.0, "epoch": 1, "t": 10.974442958831787}
|
gemma-3-27b-it/haiku_bottom-s0/README.md
ADDED
|
@@ -0,0 +1,206 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: google/gemma-3-27b-it
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- base_model:adapter:google/gemma-3-27b-it
|
| 6 |
+
- lora
|
| 7 |
+
- transformers
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# Model Card for Model ID
|
| 11 |
+
|
| 12 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
## Model Details
|
| 17 |
+
|
| 18 |
+
### Model Description
|
| 19 |
+
|
| 20 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
- **Developed by:** [More Information Needed]
|
| 25 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 26 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 27 |
+
- **Model type:** [More Information Needed]
|
| 28 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 29 |
+
- **License:** [More Information Needed]
|
| 30 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 31 |
+
|
| 32 |
+
### Model Sources [optional]
|
| 33 |
+
|
| 34 |
+
<!-- Provide the basic links for the model. -->
|
| 35 |
+
|
| 36 |
+
- **Repository:** [More Information Needed]
|
| 37 |
+
- **Paper [optional]:** [More Information Needed]
|
| 38 |
+
- **Demo [optional]:** [More Information Needed]
|
| 39 |
+
|
| 40 |
+
## Uses
|
| 41 |
+
|
| 42 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 43 |
+
|
| 44 |
+
### Direct Use
|
| 45 |
+
|
| 46 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 47 |
+
|
| 48 |
+
[More Information Needed]
|
| 49 |
+
|
| 50 |
+
### Downstream Use [optional]
|
| 51 |
+
|
| 52 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 53 |
+
|
| 54 |
+
[More Information Needed]
|
| 55 |
+
|
| 56 |
+
### Out-of-Scope Use
|
| 57 |
+
|
| 58 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 59 |
+
|
| 60 |
+
[More Information Needed]
|
| 61 |
+
|
| 62 |
+
## Bias, Risks, and Limitations
|
| 63 |
+
|
| 64 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 65 |
+
|
| 66 |
+
[More Information Needed]
|
| 67 |
+
|
| 68 |
+
### Recommendations
|
| 69 |
+
|
| 70 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 71 |
+
|
| 72 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 73 |
+
|
| 74 |
+
## How to Get Started with the Model
|
| 75 |
+
|
| 76 |
+
Use the code below to get started with the model.
|
| 77 |
+
|
| 78 |
+
[More Information Needed]
|
| 79 |
+
|
| 80 |
+
## Training Details
|
| 81 |
+
|
| 82 |
+
### Training Data
|
| 83 |
+
|
| 84 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 85 |
+
|
| 86 |
+
[More Information Needed]
|
| 87 |
+
|
| 88 |
+
### Training Procedure
|
| 89 |
+
|
| 90 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 91 |
+
|
| 92 |
+
#### Preprocessing [optional]
|
| 93 |
+
|
| 94 |
+
[More Information Needed]
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
#### Training Hyperparameters
|
| 98 |
+
|
| 99 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 100 |
+
|
| 101 |
+
#### Speeds, Sizes, Times [optional]
|
| 102 |
+
|
| 103 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 104 |
+
|
| 105 |
+
[More Information Needed]
|
| 106 |
+
|
| 107 |
+
## Evaluation
|
| 108 |
+
|
| 109 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 110 |
+
|
| 111 |
+
### Testing Data, Factors & Metrics
|
| 112 |
+
|
| 113 |
+
#### Testing Data
|
| 114 |
+
|
| 115 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 116 |
+
|
| 117 |
+
[More Information Needed]
|
| 118 |
+
|
| 119 |
+
#### Factors
|
| 120 |
+
|
| 121 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 122 |
+
|
| 123 |
+
[More Information Needed]
|
| 124 |
+
|
| 125 |
+
#### Metrics
|
| 126 |
+
|
| 127 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 128 |
+
|
| 129 |
+
[More Information Needed]
|
| 130 |
+
|
| 131 |
+
### Results
|
| 132 |
+
|
| 133 |
+
[More Information Needed]
|
| 134 |
+
|
| 135 |
+
#### Summary
|
| 136 |
+
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
## Model Examination [optional]
|
| 140 |
+
|
| 141 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 142 |
+
|
| 143 |
+
[More Information Needed]
|
| 144 |
+
|
| 145 |
+
## Environmental Impact
|
| 146 |
+
|
| 147 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 148 |
+
|
| 149 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 150 |
+
|
| 151 |
+
- **Hardware Type:** [More Information Needed]
|
| 152 |
+
- **Hours used:** [More Information Needed]
|
| 153 |
+
- **Cloud Provider:** [More Information Needed]
|
| 154 |
+
- **Compute Region:** [More Information Needed]
|
| 155 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 156 |
+
|
| 157 |
+
## Technical Specifications [optional]
|
| 158 |
+
|
| 159 |
+
### Model Architecture and Objective
|
| 160 |
+
|
| 161 |
+
[More Information Needed]
|
| 162 |
+
|
| 163 |
+
### Compute Infrastructure
|
| 164 |
+
|
| 165 |
+
[More Information Needed]
|
| 166 |
+
|
| 167 |
+
#### Hardware
|
| 168 |
+
|
| 169 |
+
[More Information Needed]
|
| 170 |
+
|
| 171 |
+
#### Software
|
| 172 |
+
|
| 173 |
+
[More Information Needed]
|
| 174 |
+
|
| 175 |
+
## Citation [optional]
|
| 176 |
+
|
| 177 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 178 |
+
|
| 179 |
+
**BibTeX:**
|
| 180 |
+
|
| 181 |
+
[More Information Needed]
|
| 182 |
+
|
| 183 |
+
**APA:**
|
| 184 |
+
|
| 185 |
+
[More Information Needed]
|
| 186 |
+
|
| 187 |
+
## Glossary [optional]
|
| 188 |
+
|
| 189 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 190 |
+
|
| 191 |
+
[More Information Needed]
|
| 192 |
+
|
| 193 |
+
## More Information [optional]
|
| 194 |
+
|
| 195 |
+
[More Information Needed]
|
| 196 |
+
|
| 197 |
+
## Model Card Authors [optional]
|
| 198 |
+
|
| 199 |
+
[More Information Needed]
|
| 200 |
+
|
| 201 |
+
## Model Card Contact
|
| 202 |
+
|
| 203 |
+
[More Information Needed]
|
| 204 |
+
### Framework versions
|
| 205 |
+
|
| 206 |
+
- PEFT 0.17.1
|
gemma-3-27b-it/haiku_bottom-s0/adapter_config.json
ADDED
|
@@ -0,0 +1,228 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": {
|
| 4 |
+
"base_model_class": "Gemma3ForConditionalGeneration",
|
| 5 |
+
"parent_library": "transformers.models.gemma3.modeling_gemma3"
|
| 6 |
+
},
|
| 7 |
+
"base_model_name_or_path": "google/gemma-3-27b-it",
|
| 8 |
+
"bias": "none",
|
| 9 |
+
"corda_config": null,
|
| 10 |
+
"eva_config": null,
|
| 11 |
+
"exclude_modules": null,
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": true,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layer_replication": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"layers_to_transform": null,
|
| 18 |
+
"loftq_config": {},
|
| 19 |
+
"lora_alpha": 128,
|
| 20 |
+
"lora_bias": false,
|
| 21 |
+
"lora_dropout": 0.05,
|
| 22 |
+
"megatron_config": null,
|
| 23 |
+
"megatron_core": "megatron.core",
|
| 24 |
+
"modules_to_save": null,
|
| 25 |
+
"peft_type": "LORA",
|
| 26 |
+
"qalora_group_size": 16,
|
| 27 |
+
"r": 64,
|
| 28 |
+
"rank_pattern": {},
|
| 29 |
+
"revision": null,
|
| 30 |
+
"target_modules": [
|
| 31 |
+
"31.self_attn.k_proj",
|
| 32 |
+
"55.self_attn.k_proj",
|
| 33 |
+
"34.self_attn.q_proj",
|
| 34 |
+
"language_model.layers.11.self_attn.k_proj",
|
| 35 |
+
"language_model.layers.6.self_attn.k_proj",
|
| 36 |
+
"60.self_attn.q_proj",
|
| 37 |
+
"language_model.layers.14.self_attn.v_proj",
|
| 38 |
+
"30.self_attn.v_proj",
|
| 39 |
+
"language_model.layers.24.self_attn.v_proj",
|
| 40 |
+
"32.self_attn.v_proj",
|
| 41 |
+
"language_model.layers.19.self_attn.v_proj",
|
| 42 |
+
"39.self_attn.q_proj",
|
| 43 |
+
"59.self_attn.k_proj",
|
| 44 |
+
"gate_proj",
|
| 45 |
+
"up_proj",
|
| 46 |
+
"language_model.layers.9.self_attn.k_proj",
|
| 47 |
+
"58.self_attn.v_proj",
|
| 48 |
+
"language_model.layers.18.self_attn.k_proj",
|
| 49 |
+
"language_model.layers.21.self_attn.k_proj",
|
| 50 |
+
"29.self_attn.q_proj",
|
| 51 |
+
"language_model.layers.4.self_attn.k_proj",
|
| 52 |
+
"33.self_attn.q_proj",
|
| 53 |
+
"57.self_attn.k_proj",
|
| 54 |
+
"language_model.layers.14.self_attn.k_proj",
|
| 55 |
+
"61.self_attn.v_proj",
|
| 56 |
+
"41.self_attn.q_proj",
|
| 57 |
+
"language_model.layers.3.self_attn.q_proj",
|
| 58 |
+
"o_proj",
|
| 59 |
+
"language_model.layers.7.self_attn.v_proj",
|
| 60 |
+
"46.self_attn.q_proj",
|
| 61 |
+
"51.self_attn.k_proj",
|
| 62 |
+
"54.self_attn.k_proj",
|
| 63 |
+
"language_model.layers.4.self_attn.q_proj",
|
| 64 |
+
"language_model.layers.6.self_attn.q_proj",
|
| 65 |
+
"32.self_attn.k_proj",
|
| 66 |
+
"language_model.layers.12.self_attn.q_proj",
|
| 67 |
+
"language_model.layers.7.self_attn.k_proj",
|
| 68 |
+
"language_model.layers.12.self_attn.v_proj",
|
| 69 |
+
"language_model.layers.3.self_attn.v_proj",
|
| 70 |
+
"language_model.layers.26.self_attn.v_proj",
|
| 71 |
+
"language_model.layers.8.self_attn.k_proj",
|
| 72 |
+
"language_model.layers.10.self_attn.k_proj",
|
| 73 |
+
"51.self_attn.v_proj",
|
| 74 |
+
"58.self_attn.k_proj",
|
| 75 |
+
"48.self_attn.q_proj",
|
| 76 |
+
"language_model.layers.25.self_attn.v_proj",
|
| 77 |
+
"language_model.layers.26.self_attn.q_proj",
|
| 78 |
+
"language_model.layers.22.self_attn.q_proj",
|
| 79 |
+
"49.self_attn.v_proj",
|
| 80 |
+
"language_model.layers.5.self_attn.k_proj",
|
| 81 |
+
"54.self_attn.q_proj",
|
| 82 |
+
"31.self_attn.v_proj",
|
| 83 |
+
"44.self_attn.v_proj",
|
| 84 |
+
"language_model.layers.26.self_attn.k_proj",
|
| 85 |
+
"language_model.layers.16.self_attn.q_proj",
|
| 86 |
+
"59.self_attn.q_proj",
|
| 87 |
+
"53.self_attn.q_proj",
|
| 88 |
+
"language_model.layers.2.self_attn.v_proj",
|
| 89 |
+
"49.self_attn.k_proj",
|
| 90 |
+
"language_model.layers.18.self_attn.q_proj",
|
| 91 |
+
"49.self_attn.q_proj",
|
| 92 |
+
"46.self_attn.k_proj",
|
| 93 |
+
"28.self_attn.q_proj",
|
| 94 |
+
"language_model.layers.10.self_attn.q_proj",
|
| 95 |
+
"language_model.layers.11.self_attn.v_proj",
|
| 96 |
+
"language_model.layers.24.self_attn.k_proj",
|
| 97 |
+
"language_model.layers.21.self_attn.v_proj",
|
| 98 |
+
"language_model.layers.5.self_attn.q_proj",
|
| 99 |
+
"33.self_attn.k_proj",
|
| 100 |
+
"language_model.layers.8.self_attn.q_proj",
|
| 101 |
+
"55.self_attn.q_proj",
|
| 102 |
+
"44.self_attn.k_proj",
|
| 103 |
+
"54.self_attn.v_proj",
|
| 104 |
+
"language_model.layers.20.self_attn.k_proj",
|
| 105 |
+
"56.self_attn.v_proj",
|
| 106 |
+
"42.self_attn.k_proj",
|
| 107 |
+
"33.self_attn.v_proj",
|
| 108 |
+
"43.self_attn.q_proj",
|
| 109 |
+
"43.self_attn.v_proj",
|
| 110 |
+
"40.self_attn.q_proj",
|
| 111 |
+
"42.self_attn.v_proj",
|
| 112 |
+
"37.self_attn.q_proj",
|
| 113 |
+
"language_model.layers.2.self_attn.k_proj",
|
| 114 |
+
"language_model.layers.13.self_attn.q_proj",
|
| 115 |
+
"language_model.layers.19.self_attn.k_proj",
|
| 116 |
+
"language_model.layers.17.self_attn.v_proj",
|
| 117 |
+
"language_model.layers.18.self_attn.v_proj",
|
| 118 |
+
"language_model.layers.0.self_attn.k_proj",
|
| 119 |
+
"language_model.layers.25.self_attn.k_proj",
|
| 120 |
+
"29.self_attn.k_proj",
|
| 121 |
+
"35.self_attn.k_proj",
|
| 122 |
+
"61.self_attn.k_proj",
|
| 123 |
+
"language_model.layers.15.self_attn.q_proj",
|
| 124 |
+
"language_model.layers.17.self_attn.q_proj",
|
| 125 |
+
"language_model.layers.1.self_attn.k_proj",
|
| 126 |
+
"34.self_attn.v_proj",
|
| 127 |
+
"language_model.layers.14.self_attn.q_proj",
|
| 128 |
+
"40.self_attn.v_proj",
|
| 129 |
+
"language_model.layers.12.self_attn.k_proj",
|
| 130 |
+
"46.self_attn.v_proj",
|
| 131 |
+
"38.self_attn.q_proj",
|
| 132 |
+
"58.self_attn.q_proj",
|
| 133 |
+
"language_model.layers.4.self_attn.v_proj",
|
| 134 |
+
"53.self_attn.v_proj",
|
| 135 |
+
"44.self_attn.q_proj",
|
| 136 |
+
"61.self_attn.q_proj",
|
| 137 |
+
"28.self_attn.k_proj",
|
| 138 |
+
"55.self_attn.v_proj",
|
| 139 |
+
"56.self_attn.k_proj",
|
| 140 |
+
"47.self_attn.q_proj",
|
| 141 |
+
"50.self_attn.v_proj",
|
| 142 |
+
"language_model.layers.5.self_attn.v_proj",
|
| 143 |
+
"37.self_attn.v_proj",
|
| 144 |
+
"language_model.layers.21.self_attn.q_proj",
|
| 145 |
+
"language_model.layers.23.self_attn.v_proj",
|
| 146 |
+
"language_model.layers.22.self_attn.k_proj",
|
| 147 |
+
"37.self_attn.k_proj",
|
| 148 |
+
"language_model.layers.25.self_attn.q_proj",
|
| 149 |
+
"52.self_attn.q_proj",
|
| 150 |
+
"41.self_attn.v_proj",
|
| 151 |
+
"60.self_attn.k_proj",
|
| 152 |
+
"language_model.layers.23.self_attn.q_proj",
|
| 153 |
+
"language_model.layers.19.self_attn.q_proj",
|
| 154 |
+
"52.self_attn.k_proj",
|
| 155 |
+
"43.self_attn.k_proj",
|
| 156 |
+
"57.self_attn.v_proj",
|
| 157 |
+
"32.self_attn.q_proj",
|
| 158 |
+
"38.self_attn.k_proj",
|
| 159 |
+
"35.self_attn.v_proj",
|
| 160 |
+
"language_model.layers.15.self_attn.k_proj",
|
| 161 |
+
"language_model.layers.23.self_attn.k_proj",
|
| 162 |
+
"52.self_attn.v_proj",
|
| 163 |
+
"34.self_attn.k_proj",
|
| 164 |
+
"35.self_attn.q_proj",
|
| 165 |
+
"language_model.layers.13.self_attn.v_proj",
|
| 166 |
+
"45.self_attn.k_proj",
|
| 167 |
+
"language_model.layers.24.self_attn.q_proj",
|
| 168 |
+
"28.self_attn.v_proj",
|
| 169 |
+
"language_model.layers.3.self_attn.k_proj",
|
| 170 |
+
"down_proj",
|
| 171 |
+
"59.self_attn.v_proj",
|
| 172 |
+
"31.self_attn.q_proj",
|
| 173 |
+
"30.self_attn.q_proj",
|
| 174 |
+
"language_model.layers.10.self_attn.v_proj",
|
| 175 |
+
"36.self_attn.q_proj",
|
| 176 |
+
"53.self_attn.k_proj",
|
| 177 |
+
"50.self_attn.k_proj",
|
| 178 |
+
"45.self_attn.v_proj",
|
| 179 |
+
"language_model.layers.9.self_attn.q_proj",
|
| 180 |
+
"39.self_attn.v_proj",
|
| 181 |
+
"language_model.layers.20.self_attn.v_proj",
|
| 182 |
+
"36.self_attn.v_proj",
|
| 183 |
+
"47.self_attn.k_proj",
|
| 184 |
+
"56.self_attn.q_proj",
|
| 185 |
+
"language_model.layers.2.self_attn.q_proj",
|
| 186 |
+
"51.self_attn.q_proj",
|
| 187 |
+
"41.self_attn.k_proj",
|
| 188 |
+
"39.self_attn.k_proj",
|
| 189 |
+
"48.self_attn.k_proj",
|
| 190 |
+
"language_model.layers.22.self_attn.v_proj",
|
| 191 |
+
"language_model.layers.9.self_attn.v_proj",
|
| 192 |
+
"48.self_attn.v_proj",
|
| 193 |
+
"language_model.layers.6.self_attn.v_proj",
|
| 194 |
+
"27.self_attn.q_proj",
|
| 195 |
+
"57.self_attn.q_proj",
|
| 196 |
+
"29.self_attn.v_proj",
|
| 197 |
+
"47.self_attn.v_proj",
|
| 198 |
+
"42.self_attn.q_proj",
|
| 199 |
+
"language_model.layers.8.self_attn.v_proj",
|
| 200 |
+
"language_model.layers.16.self_attn.k_proj",
|
| 201 |
+
"38.self_attn.v_proj",
|
| 202 |
+
"language_model.layers.1.self_attn.q_proj",
|
| 203 |
+
"40.self_attn.k_proj",
|
| 204 |
+
"language_model.layers.11.self_attn.q_proj",
|
| 205 |
+
"language_model.layers.0.self_attn.v_proj",
|
| 206 |
+
"60.self_attn.v_proj",
|
| 207 |
+
"50.self_attn.q_proj",
|
| 208 |
+
"language_model.layers.13.self_attn.k_proj",
|
| 209 |
+
"language_model.layers.16.self_attn.v_proj",
|
| 210 |
+
"language_model.layers.1.self_attn.v_proj",
|
| 211 |
+
"27.self_attn.v_proj",
|
| 212 |
+
"language_model.layers.20.self_attn.q_proj",
|
| 213 |
+
"language_model.layers.7.self_attn.q_proj",
|
| 214 |
+
"36.self_attn.k_proj",
|
| 215 |
+
"45.self_attn.q_proj",
|
| 216 |
+
"language_model.layers.0.self_attn.q_proj",
|
| 217 |
+
"language_model.layers.15.self_attn.v_proj",
|
| 218 |
+
"language_model.layers.17.self_attn.k_proj",
|
| 219 |
+
"27.self_attn.k_proj",
|
| 220 |
+
"30.self_attn.k_proj"
|
| 221 |
+
],
|
| 222 |
+
"target_parameters": null,
|
| 223 |
+
"task_type": null,
|
| 224 |
+
"trainable_token_indices": null,
|
| 225 |
+
"use_dora": false,
|
| 226 |
+
"use_qalora": false,
|
| 227 |
+
"use_rslora": false
|
| 228 |
+
}
|
gemma-3-27b-it/haiku_bottom-s0/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7c1632421ef0e6459617ecf3c3dcb85f13e85fabe0539dd4ae9d06f985110f42
|
| 3 |
+
size 1816396048
|
gemma-3-27b-it/haiku_bottom-s0/config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "google/gemma-3-27b-it",
|
| 3 |
+
"data": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma/arms/TASKW-haiku_bottom-gemma.jsonl",
|
| 4 |
+
"run": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma/runs/haiku_bottom-s0",
|
| 5 |
+
"steps": 0,
|
| 6 |
+
"batch": 16,
|
| 7 |
+
"micro": 1,
|
| 8 |
+
"seq_len": 4096,
|
| 9 |
+
"lr": 0.0001,
|
| 10 |
+
"warmup": 10,
|
| 11 |
+
"lora_rank": 64,
|
| 12 |
+
"lora_targets": "all-linear",
|
| 13 |
+
"lora_targets_file": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/code/gemma3_lora_targets.json",
|
| 14 |
+
"seed": 0,
|
| 15 |
+
"bos": true,
|
| 16 |
+
"no_merge": false,
|
| 17 |
+
"save_at": "",
|
| 18 |
+
"replay": "",
|
| 19 |
+
"replay_frac": 0.0,
|
| 20 |
+
"constant_lr": false,
|
| 21 |
+
"schedule": "hf-cosine",
|
| 22 |
+
"min_lr_frac": 0.0,
|
| 23 |
+
"warmup_ratio": 0.05,
|
| 24 |
+
"lora_dropout": 0.05,
|
| 25 |
+
"no_pack": true,
|
| 26 |
+
"chat_sft": true,
|
| 27 |
+
"data_frac": 1.0,
|
| 28 |
+
"base_adapter": "",
|
| 29 |
+
"ckpt_every": 0,
|
| 30 |
+
"kill_after": 0,
|
| 31 |
+
"method": "doc_lm_lora",
|
| 32 |
+
"world": 2
|
| 33 |
+
}
|
gemma-3-27b-it/haiku_bottom-s0/log.jsonl
ADDED
|
@@ -0,0 +1,312 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 1, "loss": 2.3182594776153564, "grad_norm": 4.27983283996582, "lr": 1.25e-05, "epoch": 1, "t": 5.1280677318573}
|
| 2 |
+
{"step": 2, "loss": 2.2391581535339355, "grad_norm": 4.061351299285889, "lr": 1.8750000000000002e-05, "epoch": 1, "t": 4.528010845184326}
|
| 3 |
+
{"step": 3, "loss": 2.276038646697998, "grad_norm": 4.052865505218506, "lr": 2.5e-05, "epoch": 1, "t": 4.50709342956543}
|
| 4 |
+
{"step": 4, "loss": 1.9198156595230103, "grad_norm": 3.0770339965820312, "lr": 3.125e-05, "epoch": 1, "t": 4.504337549209595}
|
| 5 |
+
{"step": 5, "loss": 1.8454433679580688, "grad_norm": 2.402545690536499, "lr": 3.7500000000000003e-05, "epoch": 1, "t": 4.062540292739868}
|
| 6 |
+
{"step": 6, "loss": 1.6746344566345215, "grad_norm": 1.9951117038726807, "lr": 4.375e-05, "epoch": 1, "t": 4.040672302246094}
|
| 7 |
+
{"step": 7, "loss": 1.463655710220337, "grad_norm": 0.8613082766532898, "lr": 5e-05, "epoch": 1, "t": 5.088658571243286}
|
| 8 |
+
{"step": 8, "loss": 1.535784125328064, "grad_norm": 0.7437063455581665, "lr": 5.6250000000000005e-05, "epoch": 1, "t": 4.553965091705322}
|
| 9 |
+
{"step": 9, "loss": 1.4724493026733398, "grad_norm": 0.7593427896499634, "lr": 6.25e-05, "epoch": 1, "t": 4.844298362731934}
|
| 10 |
+
{"step": 10, "loss": 1.4228730201721191, "grad_norm": 0.8455522656440735, "lr": 6.875e-05, "epoch": 1, "t": 4.826720476150513}
|
| 11 |
+
{"step": 11, "loss": 1.3299033641815186, "grad_norm": 0.8137816190719604, "lr": 7.500000000000001e-05, "epoch": 1, "t": 4.302637100219727}
|
| 12 |
+
{"step": 12, "loss": 1.461927890777588, "grad_norm": 0.7387231588363647, "lr": 8.125000000000001e-05, "epoch": 1, "t": 5.466124057769775}
|
| 13 |
+
{"step": 13, "loss": 1.2449781894683838, "grad_norm": 0.6877515316009521, "lr": 8.75e-05, "epoch": 1, "t": 4.260915040969849}
|
| 14 |
+
{"step": 14, "loss": 1.2755296230316162, "grad_norm": 0.638268232345581, "lr": 9.375e-05, "epoch": 1, "t": 4.409672260284424}
|
| 15 |
+
{"step": 15, "loss": 1.3630908727645874, "grad_norm": 0.6184887290000916, "lr": 0.0001, "epoch": 1, "t": 4.528217077255249}
|
| 16 |
+
{"step": 16, "loss": 1.306757926940918, "grad_norm": 0.6700087189674377, "lr": 0.0001, "epoch": 1, "t": 4.423699617385864}
|
| 17 |
+
{"step": 17, "loss": 1.352484941482544, "grad_norm": 0.6894152164459229, "lr": 9.99971838728789e-05, "epoch": 1, "t": 4.5501708984375}
|
| 18 |
+
{"step": 18, "loss": 1.307474136352539, "grad_norm": 0.6101863980293274, "lr": 9.998873580873848e-05, "epoch": 1, "t": 4.684481859207153}
|
| 19 |
+
{"step": 19, "loss": 1.2638506889343262, "grad_norm": 0.5963188409805298, "lr": 9.997465675921163e-05, "epoch": 1, "t": 4.298139810562134}
|
| 20 |
+
{"step": 20, "loss": 1.274450659751892, "grad_norm": 0.5659856796264648, "lr": 9.995494831023409e-05, "epoch": 1, "t": 4.341984748840332}
|
| 21 |
+
{"step": 21, "loss": 1.2542705535888672, "grad_norm": 0.535892128944397, "lr": 9.992961268186573e-05, "epoch": 1, "t": 4.5843119621276855}
|
| 22 |
+
{"step": 22, "loss": 1.226963758468628, "grad_norm": 0.5109335780143738, "lr": 9.989865272804063e-05, "epoch": 1, "t": 5.0320351123809814}
|
| 23 |
+
{"step": 23, "loss": 1.2917513847351074, "grad_norm": 0.5298612713813782, "lr": 9.986207193624536e-05, "epoch": 1, "t": 4.843608140945435}
|
| 24 |
+
{"step": 24, "loss": 1.29721999168396, "grad_norm": 0.5348758697509766, "lr": 9.981987442712633e-05, "epoch": 1, "t": 5.082450866699219}
|
| 25 |
+
{"step": 25, "loss": 1.2361458539962769, "grad_norm": 0.5346004962921143, "lr": 9.977206495402554e-05, "epoch": 1, "t": 4.976636171340942}
|
| 26 |
+
{"step": 26, "loss": 1.269109845161438, "grad_norm": 0.5565136671066284, "lr": 9.971864890244513e-05, "epoch": 1, "t": 4.9557411670684814}
|
| 27 |
+
{"step": 27, "loss": 1.205772876739502, "grad_norm": 0.5136733055114746, "lr": 9.965963228944078e-05, "epoch": 1, "t": 4.892717123031616}
|
| 28 |
+
{"step": 28, "loss": 1.1230697631835938, "grad_norm": 0.5052443146705627, "lr": 9.959502176294383e-05, "epoch": 1, "t": 4.674621820449829}
|
| 29 |
+
{"step": 29, "loss": 1.107112169265747, "grad_norm": 0.5233135223388672, "lr": 9.95248246010126e-05, "epoch": 1, "t": 4.344412803649902}
|
| 30 |
+
{"step": 30, "loss": 1.2090479135513306, "grad_norm": 0.5024049878120422, "lr": 9.944904871101228e-05, "epoch": 1, "t": 5.341161727905273}
|
| 31 |
+
{"step": 31, "loss": 1.255561113357544, "grad_norm": 0.5716925263404846, "lr": 9.936770262872443e-05, "epoch": 1, "t": 4.527912139892578}
|
| 32 |
+
{"step": 32, "loss": 1.174899697303772, "grad_norm": 0.5564167499542236, "lr": 9.928079551738543e-05, "epoch": 1, "t": 4.685332536697388}
|
| 33 |
+
{"step": 33, "loss": 1.1594640016555786, "grad_norm": 0.586647629737854, "lr": 9.918833716665419e-05, "epoch": 1, "t": 4.7574989795684814}
|
| 34 |
+
{"step": 34, "loss": 1.2194890975952148, "grad_norm": 0.6168203353881836, "lr": 9.909033799150946e-05, "epoch": 1, "t": 4.320326566696167}
|
| 35 |
+
{"step": 35, "loss": 1.1753389835357666, "grad_norm": 0.6045057773590088, "lr": 9.898680903107666e-05, "epoch": 1, "t": 4.5131001472473145}
|
| 36 |
+
{"step": 36, "loss": 1.1112675666809082, "grad_norm": 0.5438430309295654, "lr": 9.887776194738432e-05, "epoch": 1, "t": 4.208852529525757}
|
| 37 |
+
{"step": 37, "loss": 0.9868287444114685, "grad_norm": 0.5492530465126038, "lr": 9.876320902405042e-05, "epoch": 1, "t": 4.571593284606934}
|
| 38 |
+
{"step": 38, "loss": 1.2119451761245728, "grad_norm": 0.5478034615516663, "lr": 9.864316316489873e-05, "epoch": 1, "t": 4.373960018157959}
|
| 39 |
+
{"step": 39, "loss": 1.1598364114761353, "grad_norm": 0.5279031991958618, "lr": 9.851763789250525e-05, "epoch": 1, "t": 4.439868688583374}
|
| 40 |
+
{"step": 40, "loss": 1.2911561727523804, "grad_norm": 0.561524510383606, "lr": 9.838664734667495e-05, "epoch": 1, "t": 4.412939786911011}
|
| 41 |
+
{"step": 41, "loss": 1.1608424186706543, "grad_norm": 0.5272015333175659, "lr": 9.825020628284896e-05, "epoch": 1, "t": 4.482903242111206}
|
| 42 |
+
{"step": 42, "loss": 1.1495981216430664, "grad_norm": 0.5450555086135864, "lr": 9.810833007044247e-05, "epoch": 1, "t": 4.218104124069214}
|
| 43 |
+
{"step": 43, "loss": 1.0959913730621338, "grad_norm": 0.5038524866104126, "lr": 9.796103469111351e-05, "epoch": 1, "t": 4.903516530990601}
|
| 44 |
+
{"step": 44, "loss": 1.0306528806686401, "grad_norm": 0.580023467540741, "lr": 9.780833673696254e-05, "epoch": 1, "t": 4.51888370513916}
|
| 45 |
+
{"step": 45, "loss": 1.094822645187378, "grad_norm": 0.5627464652061462, "lr": 9.76502534086636e-05, "epoch": 1, "t": 4.369114875793457}
|
| 46 |
+
{"step": 46, "loss": 1.0199453830718994, "grad_norm": 0.5088703632354736, "lr": 9.74868025135266e-05, "epoch": 1, "t": 4.441473722457886}
|
| 47 |
+
{"step": 47, "loss": 1.2840280532836914, "grad_norm": 0.5627360343933105, "lr": 9.731800246349148e-05, "epoch": 1, "t": 4.7004714012146}
|
| 48 |
+
{"step": 48, "loss": 1.1457535028457642, "grad_norm": 0.5226135849952698, "lr": 9.714387227305422e-05, "epoch": 1, "t": 4.882219552993774}
|
| 49 |
+
{"step": 49, "loss": 1.1662487983703613, "grad_norm": 0.533386766910553, "lr": 9.696443155712486e-05, "epoch": 1, "t": 4.557215690612793}
|
| 50 |
+
{"step": 50, "loss": 1.1086162328720093, "grad_norm": 0.5301807522773743, "lr": 9.67797005288181e-05, "epoch": 1, "t": 4.753899574279785}
|
| 51 |
+
{"step": 51, "loss": 1.1031875610351562, "grad_norm": 0.6416805982589722, "lr": 9.65896999971763e-05, "epoch": 1, "t": 3.944395065307617}
|
| 52 |
+
{"step": 52, "loss": 1.1556980609893799, "grad_norm": 0.5035192966461182, "lr": 9.639445136482548e-05, "epoch": 1, "t": 4.874272346496582}
|
| 53 |
+
{"step": 53, "loss": 1.0844943523406982, "grad_norm": 0.6320053935050964, "lr": 9.619397662556435e-05, "epoch": 1, "t": 3.943636894226074}
|
| 54 |
+
{"step": 54, "loss": 1.1466505527496338, "grad_norm": 0.5937464833259583, "lr": 9.598829836188694e-05, "epoch": 1, "t": 4.185172080993652}
|
| 55 |
+
{"step": 55, "loss": 1.169416069984436, "grad_norm": 0.5557771921157837, "lr": 9.577743974243874e-05, "epoch": 1, "t": 4.230347633361816}
|
| 56 |
+
{"step": 56, "loss": 1.1469193696975708, "grad_norm": 0.5882948637008667, "lr": 9.55614245194068e-05, "epoch": 1, "t": 4.496831655502319}
|
| 57 |
+
{"step": 57, "loss": 1.2191901206970215, "grad_norm": 0.5372564196586609, "lr": 9.534027702584425e-05, "epoch": 1, "t": 4.598284959793091}
|
| 58 |
+
{"step": 58, "loss": 1.1269330978393555, "grad_norm": 0.5687738060951233, "lr": 9.511402217292926e-05, "epoch": 1, "t": 4.195266962051392}
|
| 59 |
+
{"step": 59, "loss": 1.2001675367355347, "grad_norm": 0.5253062844276428, "lr": 9.488268544715896e-05, "epoch": 1, "t": 5.253775358200073}
|
| 60 |
+
{"step": 60, "loss": 1.059751272201538, "grad_norm": 0.5007998943328857, "lr": 9.464629290747842e-05, "epoch": 1, "t": 4.73201847076416}
|
| 61 |
+
{"step": 61, "loss": 1.0079600811004639, "grad_norm": 0.5089836716651917, "lr": 9.440487118234535e-05, "epoch": 1, "t": 4.171860456466675}
|
| 62 |
+
{"step": 62, "loss": 1.0606391429901123, "grad_norm": 0.5798132419586182, "lr": 9.415844746673047e-05, "epoch": 1, "t": 4.0498621463775635}
|
| 63 |
+
{"step": 63, "loss": 1.1056632995605469, "grad_norm": 0.5342990159988403, "lr": 9.390704951905411e-05, "epoch": 1, "t": 4.358034133911133}
|
| 64 |
+
{"step": 64, "loss": 1.1974354982376099, "grad_norm": 0.6184249520301819, "lr": 9.365070565805941e-05, "epoch": 1, "t": 4.097787141799927}
|
| 65 |
+
{"step": 65, "loss": 1.238830327987671, "grad_norm": 0.5231149792671204, "lr": 9.338944475962237e-05, "epoch": 1, "t": 4.863854646682739}
|
| 66 |
+
{"step": 66, "loss": 1.1731104850769043, "grad_norm": 0.5707106590270996, "lr": 9.312329625349902e-05, "epoch": 1, "t": 4.195391893386841}
|
| 67 |
+
{"step": 67, "loss": 1.0703424215316772, "grad_norm": 0.5170727968215942, "lr": 9.285229012001047e-05, "epoch": 1, "t": 4.462587594985962}
|
| 68 |
+
{"step": 68, "loss": 0.9579602479934692, "grad_norm": 0.5378182530403137, "lr": 9.257645688666556e-05, "epoch": 1, "t": 4.242594003677368}
|
| 69 |
+
{"step": 69, "loss": 1.1021186113357544, "grad_norm": 0.6047465801239014, "lr": 9.22958276247223e-05, "epoch": 1, "t": 4.46303653717041}
|
| 70 |
+
{"step": 70, "loss": 1.2138588428497314, "grad_norm": 0.5210340023040771, "lr": 9.201043394568773e-05, "epoch": 1, "t": 5.061203241348267}
|
| 71 |
+
{"step": 71, "loss": 1.281541347503662, "grad_norm": 0.589238703250885, "lr": 9.172030799775699e-05, "epoch": 1, "t": 4.427860736846924}
|
| 72 |
+
{"step": 72, "loss": 1.0390024185180664, "grad_norm": 0.5161624550819397, "lr": 9.142548246219212e-05, "epoch": 1, "t": 4.686448812484741}
|
| 73 |
+
{"step": 73, "loss": 0.9110280275344849, "grad_norm": 0.5641354918479919, "lr": 9.112599054964057e-05, "epoch": 1, "t": 4.032199859619141}
|
| 74 |
+
{"step": 74, "loss": 1.2140576839447021, "grad_norm": 0.506525456905365, "lr": 9.082186599639428e-05, "epoch": 1, "t": 4.954009056091309}
|
| 75 |
+
{"step": 75, "loss": 1.1521708965301514, "grad_norm": 0.5057822465896606, "lr": 9.051314306058933e-05, "epoch": 1, "t": 4.658722162246704}
|
| 76 |
+
{"step": 76, "loss": 1.1215031147003174, "grad_norm": 0.48477110266685486, "lr": 9.019985651834703e-05, "epoch": 1, "t": 4.488047361373901}
|
| 77 |
+
{"step": 77, "loss": 1.0814063549041748, "grad_norm": 0.508513331413269, "lr": 8.988204165985649e-05, "epoch": 1, "t": 4.154196500778198}
|
| 78 |
+
{"step": 78, "loss": 1.098646879196167, "grad_norm": 0.5605230331420898, "lr": 8.955973428539944e-05, "epoch": 1, "t": 4.091546297073364}
|
| 79 |
+
{"step": 79, "loss": 1.0545575618743896, "grad_norm": 0.4966557025909424, "lr": 8.923297070131737e-05, "epoch": 1, "t": 4.371098041534424}
|
| 80 |
+
{"step": 80, "loss": 1.2122673988342285, "grad_norm": 0.5091938376426697, "lr": 8.890178771592199e-05, "epoch": 1, "t": 4.502809286117554}
|
| 81 |
+
{"step": 81, "loss": 1.0079820156097412, "grad_norm": 0.6124518513679504, "lr": 8.856622263534875e-05, "epoch": 1, "t": 4.4716479778289795}
|
| 82 |
+
{"step": 82, "loss": 1.084323763847351, "grad_norm": 0.6137211918830872, "lr": 8.822631325935463e-05, "epoch": 1, "t": 4.10309100151062}
|
| 83 |
+
{"step": 83, "loss": 1.0084223747253418, "grad_norm": 0.5893185138702393, "lr": 8.788209787706015e-05, "epoch": 1, "t": 3.7443480491638184}
|
| 84 |
+
{"step": 84, "loss": 1.0328024625778198, "grad_norm": 0.7217421531677246, "lr": 8.753361526263621e-05, "epoch": 1, "t": 5.0470075607299805}
|
| 85 |
+
{"step": 85, "loss": 1.0865235328674316, "grad_norm": 0.5847263336181641, "lr": 8.718090467093654e-05, "epoch": 1, "t": 4.688053369522095}
|
| 86 |
+
{"step": 86, "loss": 1.0836131572723389, "grad_norm": 0.9780337810516357, "lr": 8.682400583307562e-05, "epoch": 1, "t": 4.2437474727630615}
|
| 87 |
+
{"step": 87, "loss": 1.0456180572509766, "grad_norm": 0.5278544425964355, "lr": 8.646295895195333e-05, "epoch": 1, "t": 4.557479381561279}
|
| 88 |
+
{"step": 88, "loss": 1.0994131565093994, "grad_norm": 0.5237377882003784, "lr": 8.609780469772623e-05, "epoch": 1, "t": 4.438060522079468}
|
| 89 |
+
{"step": 89, "loss": 1.1217817068099976, "grad_norm": 0.56512850522995, "lr": 8.572858420322627e-05, "epoch": 1, "t": 4.071814060211182}
|
| 90 |
+
{"step": 90, "loss": 1.1419236660003662, "grad_norm": 0.5304071307182312, "lr": 8.535533905932738e-05, "epoch": 1, "t": 4.36553168296814}
|
| 91 |
+
{"step": 91, "loss": 1.0185911655426025, "grad_norm": 0.48035383224487305, "lr": 8.497811131026046e-05, "epoch": 1, "t": 4.525289058685303}
|
| 92 |
+
{"step": 92, "loss": 1.0598928928375244, "grad_norm": 0.5172107219696045, "lr": 8.459694344887732e-05, "epoch": 1, "t": 4.6219446659088135}
|
| 93 |
+
{"step": 93, "loss": 1.161812424659729, "grad_norm": 0.5744761824607849, "lr": 8.421187841186402e-05, "epoch": 1, "t": 4.073018789291382}
|
| 94 |
+
{"step": 94, "loss": 1.0621559619903564, "grad_norm": 0.5731527209281921, "lr": 8.382295957490436e-05, "epoch": 1, "t": 4.601446628570557}
|
| 95 |
+
{"step": 95, "loss": 1.1462032794952393, "grad_norm": 0.5267546772956848, "lr": 8.343023074779368e-05, "epoch": 1, "t": 4.446707010269165}
|
| 96 |
+
{"step": 96, "loss": 1.020163655281067, "grad_norm": 0.5306865572929382, "lr": 8.303373616950408e-05, "epoch": 1, "t": 4.762771368026733}
|
| 97 |
+
{"step": 97, "loss": 1.1626460552215576, "grad_norm": 0.5645160675048828, "lr": 8.263352050320094e-05, "epoch": 1, "t": 4.449122905731201}
|
| 98 |
+
{"step": 98, "loss": 1.0628409385681152, "grad_norm": 0.513582706451416, "lr": 8.222962883121196e-05, "epoch": 1, "t": 4.577300071716309}
|
| 99 |
+
{"step": 99, "loss": 1.0532479286193848, "grad_norm": 0.5487250685691833, "lr": 8.182210664994878e-05, "epoch": 1, "t": 4.386420965194702}
|
| 100 |
+
{"step": 100, "loss": 0.9273568391799927, "grad_norm": 0.5190582275390625, "lr": 8.141099986478212e-05, "epoch": 1, "t": 4.387535333633423}
|
| 101 |
+
{"step": 101, "loss": 1.1057411432266235, "grad_norm": 0.4992278516292572, "lr": 8.099635478487064e-05, "epoch": 1, "t": 4.819383144378662}
|
| 102 |
+
{"step": 102, "loss": 1.123875379562378, "grad_norm": 0.5524612665176392, "lr": 8.057821811794458e-05, "epoch": 1, "t": 4.398582220077515}
|
| 103 |
+
{"step": 103, "loss": 1.0887521505355835, "grad_norm": 0.565819263458252, "lr": 8.015663696504422e-05, "epoch": 1, "t": 4.442178010940552}
|
| 104 |
+
{"step": 104, "loss": 1.054804801940918, "grad_norm": 0.5269464254379272, "lr": 7.973165881521434e-05, "epoch": 1, "t": 4.3145434856414795}
|
| 105 |
+
{"step": 105, "loss": 1.1132597923278809, "grad_norm": 0.5262883305549622, "lr": 7.930333154015466e-05, "epoch": 1, "t": 4.597621440887451}
|
| 106 |
+
{"step": 106, "loss": 1.1616182327270508, "grad_norm": 0.5879294276237488, "lr": 7.88717033888274e-05, "epoch": 1, "t": 3.9895129203796387}
|
| 107 |
+
{"step": 107, "loss": 1.047626256942749, "grad_norm": 0.5814321637153625, "lr": 7.843682298202235e-05, "epoch": 1, "t": 4.23612642288208}
|
| 108 |
+
{"step": 108, "loss": 1.144451379776001, "grad_norm": 0.5584962368011475, "lr": 7.799873930687978e-05, "epoch": 1, "t": 4.495697736740112}
|
| 109 |
+
{"step": 109, "loss": 1.026404619216919, "grad_norm": 0.5226858854293823, "lr": 7.755750171137246e-05, "epoch": 1, "t": 4.320061683654785}
|
| 110 |
+
{"step": 110, "loss": 1.0900294780731201, "grad_norm": 0.53071528673172, "lr": 7.711315989874677e-05, "epoch": 1, "t": 4.577735185623169}
|
| 111 |
+
{"step": 111, "loss": 1.0614886283874512, "grad_norm": 0.608755886554718, "lr": 7.666576392192389e-05, "epoch": 1, "t": 3.928135395050049}
|
| 112 |
+
{"step": 112, "loss": 1.106485366821289, "grad_norm": 0.5223899483680725, "lr": 7.621536417786159e-05, "epoch": 1, "t": 4.3053624629974365}
|
| 113 |
+
{"step": 113, "loss": 1.1297920942306519, "grad_norm": 0.5254682302474976, "lr": 7.576201140187727e-05, "epoch": 1, "t": 5.031718730926514}
|
| 114 |
+
{"step": 114, "loss": 1.0280169248580933, "grad_norm": 0.5055007338523865, "lr": 7.530575666193283e-05, "epoch": 1, "t": 4.56247091293335}
|
| 115 |
+
{"step": 115, "loss": 1.1302471160888672, "grad_norm": 0.5785984992980957, "lr": 7.484665135288213e-05, "epoch": 1, "t": 4.644721031188965}
|
| 116 |
+
{"step": 116, "loss": 1.120363712310791, "grad_norm": 0.5624465346336365, "lr": 7.438474719068173e-05, "epoch": 1, "t": 4.270162343978882}
|
| 117 |
+
{"step": 117, "loss": 1.0795762538909912, "grad_norm": 0.51219242811203, "lr": 7.392009620656513e-05, "epoch": 1, "t": 4.882677316665649}
|
| 118 |
+
{"step": 118, "loss": 1.1195564270019531, "grad_norm": 0.5691790580749512, "lr": 7.345275074118185e-05, "epoch": 1, "t": 4.51539158821106}
|
| 119 |
+
{"step": 119, "loss": 1.0504623651504517, "grad_norm": 0.557971715927124, "lr": 7.298276343870151e-05, "epoch": 1, "t": 4.341449737548828}
|
| 120 |
+
{"step": 120, "loss": 1.022754192352295, "grad_norm": 0.5085095763206482, "lr": 7.251018724088367e-05, "epoch": 1, "t": 4.581089019775391}
|
| 121 |
+
{"step": 121, "loss": 1.1930490732192993, "grad_norm": 0.5402499437332153, "lr": 7.203507538111423e-05, "epoch": 1, "t": 4.29120945930481}
|
| 122 |
+
{"step": 122, "loss": 0.9849798083305359, "grad_norm": 0.4865000545978546, "lr": 7.155748137840892e-05, "epoch": 1, "t": 4.833763837814331}
|
| 123 |
+
{"step": 123, "loss": 1.0634889602661133, "grad_norm": 0.5702819228172302, "lr": 7.107745903138472e-05, "epoch": 1, "t": 3.9815304279327393}
|
| 124 |
+
{"step": 124, "loss": 1.0977647304534912, "grad_norm": 0.5151114463806152, "lr": 7.059506241219965e-05, "epoch": 1, "t": 5.006288051605225}
|
| 125 |
+
{"step": 125, "loss": 0.9465334415435791, "grad_norm": 0.601161777973175, "lr": 7.011034586046176e-05, "epoch": 1, "t": 3.9056437015533447}
|
| 126 |
+
{"step": 126, "loss": 1.1422866582870483, "grad_norm": 0.5249250531196594, "lr": 6.962336397710819e-05, "epoch": 1, "t": 4.667974948883057}
|
| 127 |
+
{"step": 127, "loss": 1.0053578615188599, "grad_norm": 0.480178564786911, "lr": 6.91341716182545e-05, "epoch": 1, "t": 4.703177213668823}
|
| 128 |
+
{"step": 128, "loss": 0.9437476396560669, "grad_norm": 0.6007505655288696, "lr": 6.864282388901544e-05, "epoch": 1, "t": 3.937471866607666}
|
| 129 |
+
{"step": 129, "loss": 0.9948303699493408, "grad_norm": 0.5376791954040527, "lr": 6.814937613729766e-05, "epoch": 1, "t": 4.37124228477478}
|
| 130 |
+
{"step": 130, "loss": 1.1242518424987793, "grad_norm": 0.5657219290733337, "lr": 6.765388394756504e-05, "epoch": 1, "t": 4.458618640899658}
|
| 131 |
+
{"step": 131, "loss": 1.0851949453353882, "grad_norm": 0.5400698781013489, "lr": 6.715640313457733e-05, "epoch": 1, "t": 4.658433437347412}
|
| 132 |
+
{"step": 132, "loss": 1.0876405239105225, "grad_norm": 0.5429879426956177, "lr": 6.665698973710288e-05, "epoch": 1, "t": 4.376655578613281}
|
| 133 |
+
{"step": 133, "loss": 1.0334701538085938, "grad_norm": 0.5221796631813049, "lr": 6.615570001160626e-05, "epoch": 1, "t": 4.503961801528931}
|
| 134 |
+
{"step": 134, "loss": 1.0760217905044556, "grad_norm": 0.6059476733207703, "lr": 6.565259042591113e-05, "epoch": 1, "t": 4.067153453826904}
|
| 135 |
+
{"step": 135, "loss": 1.0673279762268066, "grad_norm": 0.5496578216552734, "lr": 6.514771765283942e-05, "epoch": 1, "t": 4.437847852706909}
|
| 136 |
+
{"step": 136, "loss": 1.123016595840454, "grad_norm": 0.5759939551353455, "lr": 6.464113856382752e-05, "epoch": 1, "t": 4.666795253753662}
|
| 137 |
+
{"step": 137, "loss": 1.092640995979309, "grad_norm": 0.5213869214057922, "lr": 6.413291022251989e-05, "epoch": 1, "t": 4.291595935821533}
|
| 138 |
+
{"step": 138, "loss": 1.164132833480835, "grad_norm": 0.5406259894371033, "lr": 6.362308987834115e-05, "epoch": 1, "t": 4.620205640792847}
|
| 139 |
+
{"step": 139, "loss": 1.030076503753662, "grad_norm": 0.5251417756080627, "lr": 6.311173496004723e-05, "epoch": 1, "t": 4.693472862243652}
|
| 140 |
+
{"step": 140, "loss": 1.1202666759490967, "grad_norm": 0.5583895444869995, "lr": 6.259890306925627e-05, "epoch": 1, "t": 4.033910751342773}
|
| 141 |
+
{"step": 141, "loss": 0.973567008972168, "grad_norm": 0.537427544593811, "lr": 6.208465197396013e-05, "epoch": 1, "t": 4.544414520263672}
|
| 142 |
+
{"step": 142, "loss": 1.018080234527588, "grad_norm": 0.5596265196800232, "lr": 6.156903960201709e-05, "epoch": 1, "t": 4.622097969055176}
|
| 143 |
+
{"step": 143, "loss": 1.0365535020828247, "grad_norm": 0.59474116563797, "lr": 6.105212403462651e-05, "epoch": 1, "t": 4.769702911376953}
|
| 144 |
+
{"step": 144, "loss": 1.123967170715332, "grad_norm": 0.47663769125938416, "lr": 6.0533963499786314e-05, "epoch": 1, "t": 5.11264181137085}
|
| 145 |
+
{"step": 145, "loss": 0.921715497970581, "grad_norm": 0.5195542573928833, "lr": 6.001461636573397e-05, "epoch": 1, "t": 4.1595940589904785}
|
| 146 |
+
{"step": 146, "loss": 1.0620217323303223, "grad_norm": 0.5452298521995544, "lr": 5.949414113437142e-05, "epoch": 1, "t": 4.677001953125}
|
| 147 |
+
{"step": 147, "loss": 1.0726895332336426, "grad_norm": 0.5352897047996521, "lr": 5.897259643467527e-05, "epoch": 1, "t": 4.307942867279053}
|
| 148 |
+
{"step": 148, "loss": 1.1086431741714478, "grad_norm": 0.5879285335540771, "lr": 5.8450041016092464e-05, "epoch": 1, "t": 4.195312261581421}
|
| 149 |
+
{"step": 149, "loss": 1.0972446203231812, "grad_norm": 0.5766710042953491, "lr": 5.792653374192245e-05, "epoch": 1, "t": 4.259337902069092}
|
| 150 |
+
{"step": 150, "loss": 1.0178098678588867, "grad_norm": 0.6017856597900391, "lr": 5.7402133582686576e-05, "epoch": 1, "t": 3.9317562580108643}
|
| 151 |
+
{"step": 151, "loss": 1.078547477722168, "grad_norm": 0.5264674425125122, "lr": 5.6876899609485256e-05, "epoch": 1, "t": 4.49289608001709}
|
| 152 |
+
{"step": 152, "loss": 1.102089524269104, "grad_norm": 0.5552437901496887, "lr": 5.6350890987343944e-05, "epoch": 1, "t": 4.442862510681152}
|
| 153 |
+
{"step": 153, "loss": 1.0655879974365234, "grad_norm": 0.4840118885040283, "lr": 5.582416696854853e-05, "epoch": 1, "t": 4.819095611572266}
|
| 154 |
+
{"step": 154, "loss": 0.918845534324646, "grad_norm": 0.5310778021812439, "lr": 5.5296786885970805e-05, "epoch": 1, "t": 4.513690710067749}
|
| 155 |
+
{"step": 155, "loss": 1.1092482805252075, "grad_norm": 0.5673016309738159, "lr": 5.476881014638491e-05, "epoch": 1, "t": 4.168788194656372}
|
| 156 |
+
{"step": 156, "loss": 1.1490557193756104, "grad_norm": 0.4792988896369934, "lr": 5.4240296223775465e-05, "epoch": 1, "t": 4.933318614959717}
|
| 157 |
+
{"step": 157, "loss": 1.0512540340423584, "grad_norm": 0.5112766623497009, "lr": 5.3711304652638126e-05, "epoch": 1, "t": 5.074463844299316}
|
| 158 |
+
{"step": 158, "loss": 1.027459740638733, "grad_norm": 0.5209557414054871, "lr": 5.318189502127332e-05, "epoch": 1, "t": 4.507224798202515}
|
| 159 |
+
{"step": 159, "loss": 1.0976512432098389, "grad_norm": 0.5071336030960083, "lr": 5.265212696507387e-05, "epoch": 1, "t": 4.511084318161011}
|
| 160 |
+
{"step": 160, "loss": 1.1377782821655273, "grad_norm": 0.48624223470687866, "lr": 5.212206015980742e-05, "epoch": 1, "t": 5.3057146072387695}
|
| 161 |
+
{"step": 161, "loss": 1.1065459251403809, "grad_norm": 0.6089529395103455, "lr": 5.159175431489424e-05, "epoch": 1, "t": 4.1306538581848145}
|
| 162 |
+
{"step": 162, "loss": 1.157946228981018, "grad_norm": 0.5385098457336426, "lr": 5.1061269166681183e-05, "epoch": 1, "t": 4.367673397064209}
|
| 163 |
+
{"step": 163, "loss": 0.9364073276519775, "grad_norm": 0.6155913472175598, "lr": 5.053066447171282e-05, "epoch": 1, "t": 3.6702613830566406}
|
| 164 |
+
{"step": 164, "loss": 1.0799977779388428, "grad_norm": 0.49823373556137085, "lr": 5e-05, "epoch": 1, "t": 4.393790006637573}
|
| 165 |
+
{"step": 165, "loss": 1.0996894836425781, "grad_norm": 0.5013274550437927, "lr": 4.94693355282872e-05, "epoch": 1, "t": 4.7591612339019775}
|
| 166 |
+
{"step": 166, "loss": 1.0865201950073242, "grad_norm": 0.5232498645782471, "lr": 4.893873083331882e-05, "epoch": 1, "t": 4.372714519500732}
|
| 167 |
+
{"step": 167, "loss": 1.0524868965148926, "grad_norm": 0.5000494122505188, "lr": 4.840824568510579e-05, "epoch": 1, "t": 4.435556411743164}
|
| 168 |
+
{"step": 168, "loss": 0.9212394952774048, "grad_norm": 0.5707553029060364, "lr": 4.78779398401926e-05, "epoch": 1, "t": 4.072383165359497}
|
| 169 |
+
{"step": 169, "loss": 1.2808607816696167, "grad_norm": 0.5759779810905457, "lr": 4.734787303492615e-05, "epoch": 1, "t": 4.717279434204102}
|
| 170 |
+
{"step": 170, "loss": 1.1311876773834229, "grad_norm": 0.4997013807296753, "lr": 4.6818104978726685e-05, "epoch": 1, "t": 4.609082460403442}
|
| 171 |
+
{"step": 171, "loss": 1.128800868988037, "grad_norm": 0.5306641459465027, "lr": 4.628869534736187e-05, "epoch": 1, "t": 4.519702434539795}
|
| 172 |
+
{"step": 172, "loss": 0.9134346842765808, "grad_norm": 0.4742140769958496, "lr": 4.575970377622456e-05, "epoch": 1, "t": 4.838714599609375}
|
| 173 |
+
{"step": 173, "loss": 1.0570604801177979, "grad_norm": 0.5552225112915039, "lr": 4.52311898536151e-05, "epoch": 1, "t": 4.4790778160095215}
|
| 174 |
+
{"step": 174, "loss": 1.019338846206665, "grad_norm": 0.5955643057823181, "lr": 4.47032131140292e-05, "epoch": 1, "t": 4.5283355712890625}
|
| 175 |
+
{"step": 175, "loss": 1.0274667739868164, "grad_norm": 0.5497900247573853, "lr": 4.4175833031451473e-05, "epoch": 1, "t": 4.457115888595581}
|
| 176 |
+
{"step": 176, "loss": 1.086936593055725, "grad_norm": 0.5342774391174316, "lr": 4.364910901265606e-05, "epoch": 1, "t": 4.374600887298584}
|
| 177 |
+
{"step": 177, "loss": 1.0455431938171387, "grad_norm": 0.5045028924942017, "lr": 4.3123100390514756e-05, "epoch": 1, "t": 4.365716934204102}
|
| 178 |
+
{"step": 178, "loss": 1.0901545286178589, "grad_norm": 0.5298687815666199, "lr": 4.2597866417313436e-05, "epoch": 1, "t": 4.253988027572632}
|
| 179 |
+
{"step": 179, "loss": 1.1432433128356934, "grad_norm": 0.5694761276245117, "lr": 4.207346625807756e-05, "epoch": 1, "t": 4.1064453125}
|
| 180 |
+
{"step": 180, "loss": 1.1150802373886108, "grad_norm": 0.5988235473632812, "lr": 4.1549958983907555e-05, "epoch": 1, "t": 4.258797883987427}
|
| 181 |
+
{"step": 181, "loss": 1.0617355108261108, "grad_norm": 0.48805877566337585, "lr": 4.102740356532473e-05, "epoch": 1, "t": 4.510316371917725}
|
| 182 |
+
{"step": 182, "loss": 1.0038782358169556, "grad_norm": 0.5585227012634277, "lr": 4.050585886562858e-05, "epoch": 1, "t": 4.691247224807739}
|
| 183 |
+
{"step": 183, "loss": 1.024353265762329, "grad_norm": 0.6235622763633728, "lr": 3.998538363426605e-05, "epoch": 1, "t": 3.9281091690063477}
|
| 184 |
+
{"step": 184, "loss": 1.1303329467773438, "grad_norm": 0.614143967628479, "lr": 3.94660365002137e-05, "epoch": 1, "t": 3.929875135421753}
|
| 185 |
+
{"step": 185, "loss": 1.0341070890426636, "grad_norm": 0.5180628299713135, "lr": 3.894787596537352e-05, "epoch": 1, "t": 5.039501428604126}
|
| 186 |
+
{"step": 186, "loss": 1.04641592502594, "grad_norm": 0.6050846576690674, "lr": 3.843096039798293e-05, "epoch": 1, "t": 3.867766857147217}
|
| 187 |
+
{"step": 187, "loss": 1.1024057865142822, "grad_norm": 0.5669552087783813, "lr": 3.791534802603988e-05, "epoch": 1, "t": 4.555077791213989}
|
| 188 |
+
{"step": 188, "loss": 1.1636323928833008, "grad_norm": 0.4995519816875458, "lr": 3.740109693074375e-05, "epoch": 1, "t": 4.734290599822998}
|
| 189 |
+
{"step": 189, "loss": 1.0735996961593628, "grad_norm": 0.5093196034431458, "lr": 3.68882650399528e-05, "epoch": 1, "t": 4.841955900192261}
|
| 190 |
+
{"step": 190, "loss": 1.0439932346343994, "grad_norm": 0.5297228693962097, "lr": 3.637691012165886e-05, "epoch": 1, "t": 4.3820412158966064}
|
| 191 |
+
{"step": 191, "loss": 1.0277173519134521, "grad_norm": 0.6291950345039368, "lr": 3.586708977748012e-05, "epoch": 1, "t": 4.358238220214844}
|
| 192 |
+
{"step": 192, "loss": 1.0819365978240967, "grad_norm": 0.5827075839042664, "lr": 3.5358861436172485e-05, "epoch": 1, "t": 4.519319295883179}
|
| 193 |
+
{"step": 193, "loss": 1.0004122257232666, "grad_norm": 0.5812299847602844, "lr": 3.485228234716058e-05, "epoch": 1, "t": 4.060560941696167}
|
| 194 |
+
{"step": 194, "loss": 1.0828702449798584, "grad_norm": 0.6025108098983765, "lr": 3.434740957408889e-05, "epoch": 1, "t": 4.175429105758667}
|
| 195 |
+
{"step": 195, "loss": 1.1582586765289307, "grad_norm": 0.49842584133148193, "lr": 3.3844299988393755e-05, "epoch": 1, "t": 5.127326250076294}
|
| 196 |
+
{"step": 196, "loss": 0.9903317093849182, "grad_norm": 0.533919632434845, "lr": 3.334301026289712e-05, "epoch": 1, "t": 4.460814952850342}
|
| 197 |
+
{"step": 197, "loss": 0.9877713918685913, "grad_norm": 0.5166082382202148, "lr": 3.284359686542269e-05, "epoch": 1, "t": 4.177352666854858}
|
| 198 |
+
{"step": 198, "loss": 1.1141719818115234, "grad_norm": 0.5050948858261108, "lr": 3.234611605243496e-05, "epoch": 1, "t": 5.06512713432312}
|
| 199 |
+
{"step": 199, "loss": 1.0266294479370117, "grad_norm": 0.5247622728347778, "lr": 3.1850623862702344e-05, "epoch": 1, "t": 4.668641090393066}
|
| 200 |
+
{"step": 200, "loss": 1.0133147239685059, "grad_norm": 0.49373194575309753, "lr": 3.135717611098458e-05, "epoch": 1, "t": 5.040018081665039}
|
| 201 |
+
{"step": 201, "loss": 1.0639673471450806, "grad_norm": 0.5062780976295471, "lr": 3.086582838174551e-05, "epoch": 1, "t": 4.643780708312988}
|
| 202 |
+
{"step": 202, "loss": 1.0242486000061035, "grad_norm": 0.8172736763954163, "lr": 3.0376636022891812e-05, "epoch": 1, "t": 4.386077404022217}
|
| 203 |
+
{"step": 203, "loss": 1.012434959411621, "grad_norm": 0.5004093050956726, "lr": 2.9889654139538246e-05, "epoch": 1, "t": 4.6061179637908936}
|
| 204 |
+
{"step": 204, "loss": 1.0407299995422363, "grad_norm": 0.5462143421173096, "lr": 2.9404937587800375e-05, "epoch": 1, "t": 4.1180195808410645}
|
| 205 |
+
{"step": 205, "loss": 0.9675360321998596, "grad_norm": 0.5693405270576477, "lr": 2.8922540968615286e-05, "epoch": 1, "t": 4.175534009933472}
|
| 206 |
+
{"step": 206, "loss": 1.07175612449646, "grad_norm": 0.541511595249176, "lr": 2.8442518621591086e-05, "epoch": 1, "t": 4.615241050720215}
|
| 207 |
+
{"step": 207, "loss": 1.1230800151824951, "grad_norm": 0.4979943335056305, "lr": 2.7964924618885778e-05, "epoch": 1, "t": 5.644282817840576}
|
| 208 |
+
{"step": 208, "loss": 1.0087041854858398, "grad_norm": 0.538437008857727, "lr": 2.748981275911633e-05, "epoch": 1, "t": 4.2667787075042725}
|
| 209 |
+
{"step": 209, "loss": 1.0306894779205322, "grad_norm": 0.542610764503479, "lr": 2.701723656129851e-05, "epoch": 1, "t": 4.277086019515991}
|
| 210 |
+
{"step": 210, "loss": 1.0517539978027344, "grad_norm": 0.6919747591018677, "lr": 2.6547249258818164e-05, "epoch": 1, "t": 4.13981556892395}
|
| 211 |
+
{"step": 211, "loss": 0.9890724420547485, "grad_norm": 0.5264641046524048, "lr": 2.607990379343489e-05, "epoch": 1, "t": 4.133480548858643}
|
| 212 |
+
{"step": 212, "loss": 1.045159935951233, "grad_norm": 0.6240392923355103, "lr": 2.5615252809318284e-05, "epoch": 1, "t": 4.225189924240112}
|
| 213 |
+
{"step": 213, "loss": 1.0057790279388428, "grad_norm": 0.6362853050231934, "lr": 2.5153348647117857e-05, "epoch": 1, "t": 4.619581460952759}
|
| 214 |
+
{"step": 214, "loss": 1.063778281211853, "grad_norm": 0.5307620763778687, "lr": 2.469424333806718e-05, "epoch": 1, "t": 4.561526298522949}
|
| 215 |
+
{"step": 215, "loss": 0.9758619070053101, "grad_norm": 0.5621666312217712, "lr": 2.4237988598122752e-05, "epoch": 1, "t": 4.512598991394043}
|
| 216 |
+
{"step": 216, "loss": 1.0140074491500854, "grad_norm": 0.5423521399497986, "lr": 2.3784635822138424e-05, "epoch": 1, "t": 4.338836908340454}
|
| 217 |
+
{"step": 217, "loss": 0.972283661365509, "grad_norm": 0.5642887949943542, "lr": 2.333423607807613e-05, "epoch": 1, "t": 4.00283145904541}
|
| 218 |
+
{"step": 218, "loss": 1.0538939237594604, "grad_norm": 0.5357561707496643, "lr": 2.288684010125325e-05, "epoch": 1, "t": 4.343484878540039}
|
| 219 |
+
{"step": 219, "loss": 1.0054504871368408, "grad_norm": 0.5771626830101013, "lr": 2.2442498288627556e-05, "epoch": 1, "t": 4.1507909297943115}
|
| 220 |
+
{"step": 220, "loss": 1.0451167821884155, "grad_norm": 0.542662501335144, "lr": 2.2001260693120233e-05, "epoch": 1, "t": 4.722820520401001}
|
| 221 |
+
{"step": 221, "loss": 1.1285812854766846, "grad_norm": 0.5216588377952576, "lr": 2.156317701797766e-05, "epoch": 1, "t": 5.069575309753418}
|
| 222 |
+
{"step": 222, "loss": 1.0853524208068848, "grad_norm": 0.7681800127029419, "lr": 2.1128296611172593e-05, "epoch": 1, "t": 5.344943523406982}
|
| 223 |
+
{"step": 223, "loss": 1.033549189567566, "grad_norm": 0.6008373498916626, "lr": 2.0696668459845355e-05, "epoch": 1, "t": 5.082601070404053}
|
| 224 |
+
{"step": 224, "loss": 1.0079463720321655, "grad_norm": 0.5472951531410217, "lr": 2.026834118478567e-05, "epoch": 1, "t": 4.886492490768433}
|
| 225 |
+
{"step": 225, "loss": 1.0790884494781494, "grad_norm": 0.5625029802322388, "lr": 1.98433630349558e-05, "epoch": 1, "t": 3.9033923149108887}
|
| 226 |
+
{"step": 226, "loss": 1.0042743682861328, "grad_norm": 0.5030620694160461, "lr": 1.9421781882055444e-05, "epoch": 1, "t": 5.301595687866211}
|
| 227 |
+
{"step": 227, "loss": 1.1190781593322754, "grad_norm": 0.5105768442153931, "lr": 1.9003645215129355e-05, "epoch": 1, "t": 4.672891616821289}
|
| 228 |
+
{"step": 228, "loss": 1.0224028825759888, "grad_norm": 0.49649953842163086, "lr": 1.858900013521788e-05, "epoch": 1, "t": 4.957526206970215}
|
| 229 |
+
{"step": 229, "loss": 1.1677991151809692, "grad_norm": 0.5419648885726929, "lr": 1.817789335005121e-05, "epoch": 1, "t": 4.478188991546631}
|
| 230 |
+
{"step": 230, "loss": 0.9765380024909973, "grad_norm": 0.5728964805603027, "lr": 1.777037116878804e-05, "epoch": 1, "t": 4.242607831954956}
|
| 231 |
+
{"step": 231, "loss": 0.9520196914672852, "grad_norm": 0.5413674712181091, "lr": 1.7366479496799077e-05, "epoch": 1, "t": 3.9896466732025146}
|
| 232 |
+
{"step": 232, "loss": 1.0150930881500244, "grad_norm": 0.5681119561195374, "lr": 1.6966263830495936e-05, "epoch": 1, "t": 4.406031847000122}
|
| 233 |
+
{"step": 233, "loss": 1.0148005485534668, "grad_norm": 0.5655505061149597, "lr": 1.656976925220633e-05, "epoch": 1, "t": 4.484252452850342}
|
| 234 |
+
{"step": 234, "loss": 1.0925610065460205, "grad_norm": 0.5731245279312134, "lr": 1.6177040425095662e-05, "epoch": 1, "t": 4.79651665687561}
|
| 235 |
+
{"step": 235, "loss": 1.0736114978790283, "grad_norm": 0.5150059461593628, "lr": 1.5788121588135975e-05, "epoch": 1, "t": 5.232740640640259}
|
| 236 |
+
{"step": 236, "loss": 1.098027229309082, "grad_norm": 0.524219810962677, "lr": 1.5403056551122697e-05, "epoch": 1, "t": 4.718247890472412}
|
| 237 |
+
{"step": 237, "loss": 1.0204665660858154, "grad_norm": 0.5716584324836731, "lr": 1.5021888689739549e-05, "epoch": 1, "t": 4.200188875198364}
|
| 238 |
+
{"step": 238, "loss": 0.8543922901153564, "grad_norm": 0.538303792476654, "lr": 1.4644660940672627e-05, "epoch": 1, "t": 4.098372459411621}
|
| 239 |
+
{"step": 239, "loss": 0.9970686435699463, "grad_norm": 0.5662669539451599, "lr": 1.427141579677374e-05, "epoch": 1, "t": 4.319905519485474}
|
| 240 |
+
{"step": 240, "loss": 1.0023155212402344, "grad_norm": 0.5340131521224976, "lr": 1.3902195302273779e-05, "epoch": 1, "t": 4.177032709121704}
|
| 241 |
+
{"step": 241, "loss": 0.9963352084159851, "grad_norm": 0.56769859790802, "lr": 1.3537041048046695e-05, "epoch": 1, "t": 4.03005313873291}
|
| 242 |
+
{"step": 242, "loss": 0.9301167726516724, "grad_norm": 0.5520228743553162, "lr": 1.3175994166924394e-05, "epoch": 1, "t": 4.708632707595825}
|
| 243 |
+
{"step": 243, "loss": 1.0148307085037231, "grad_norm": 0.4941009283065796, "lr": 1.2819095329063469e-05, "epoch": 1, "t": 4.69291615486145}
|
| 244 |
+
{"step": 244, "loss": 1.0223767757415771, "grad_norm": 0.5227289795875549, "lr": 1.246638473736378e-05, "epoch": 1, "t": 4.218645334243774}
|
| 245 |
+
{"step": 245, "loss": 1.0138245820999146, "grad_norm": 0.6000615954399109, "lr": 1.2117902122939861e-05, "epoch": 1, "t": 3.7534725666046143}
|
| 246 |
+
{"step": 246, "loss": 1.1209429502487183, "grad_norm": 0.5765858292579651, "lr": 1.1773686740645384e-05, "epoch": 1, "t": 4.109350204467773}
|
| 247 |
+
{"step": 247, "loss": 1.0544321537017822, "grad_norm": 0.6045498251914978, "lr": 1.1433777364651271e-05, "epoch": 1, "t": 4.278370141983032}
|
| 248 |
+
{"step": 248, "loss": 1.0104535818099976, "grad_norm": 0.5245023965835571, "lr": 1.1098212284078036e-05, "epoch": 1, "t": 4.8333704471588135}
|
| 249 |
+
{"step": 249, "loss": 0.9180708527565002, "grad_norm": 0.5768724679946899, "lr": 1.076702929868264e-05, "epoch": 1, "t": 4.094990968704224}
|
| 250 |
+
{"step": 250, "loss": 1.1898705959320068, "grad_norm": 0.4956952631473541, "lr": 1.0440265714600572e-05, "epoch": 1, "t": 4.909663438796997}
|
| 251 |
+
{"step": 251, "loss": 1.1065311431884766, "grad_norm": 0.559075653553009, "lr": 1.0117958340143507e-05, "epoch": 1, "t": 4.577718496322632}
|
| 252 |
+
{"step": 252, "loss": 0.9776133894920349, "grad_norm": 0.5778295397758484, "lr": 9.800143481652979e-06, "epoch": 1, "t": 4.062043905258179}
|
| 253 |
+
{"step": 253, "loss": 1.1803488731384277, "grad_norm": 0.8529964089393616, "lr": 9.48685693941067e-06, "epoch": 1, "t": 4.498236417770386}
|
| 254 |
+
{"step": 254, "loss": 1.0039808750152588, "grad_norm": 0.5118018984794617, "lr": 9.17813400360572e-06, "epoch": 1, "t": 4.851653337478638}
|
| 255 |
+
{"step": 255, "loss": 1.109760046005249, "grad_norm": 0.5212645530700684, "lr": 8.874009450359427e-06, "epoch": 1, "t": 4.860409736633301}
|
| 256 |
+
{"step": 256, "loss": 1.0329498052597046, "grad_norm": 0.6550596356391907, "lr": 8.574517537807897e-06, "epoch": 1, "t": 4.919097661972046}
|
| 257 |
+
{"step": 257, "loss": 1.0815367698669434, "grad_norm": 0.5947718620300293, "lr": 8.279692002243027e-06, "epoch": 1, "t": 4.545341491699219}
|
| 258 |
+
{"step": 258, "loss": 1.0716111660003662, "grad_norm": 0.5768417119979858, "lr": 7.989566054312287e-06, "epoch": 1, "t": 4.85042142868042}
|
| 259 |
+
{"step": 259, "loss": 1.0296393632888794, "grad_norm": 0.5635685920715332, "lr": 7.704172375277691e-06, "epoch": 1, "t": 4.746562242507935}
|
| 260 |
+
{"step": 260, "loss": 1.118285059928894, "grad_norm": 0.5632694363594055, "lr": 7.423543113334436e-06, "epoch": 1, "t": 4.605138540267944}
|
| 261 |
+
{"step": 261, "loss": 1.0031137466430664, "grad_norm": 0.6423073410987854, "lr": 7.14770987998954e-06, "epoch": 1, "t": 4.276600360870361}
|
| 262 |
+
{"step": 262, "loss": 0.8432674407958984, "grad_norm": 0.6431649923324585, "lr": 6.876703746500984e-06, "epoch": 1, "t": 3.690114736557007}
|
| 263 |
+
{"step": 263, "loss": 1.100170612335205, "grad_norm": 0.5560281276702881, "lr": 6.610555240377652e-06, "epoch": 1, "t": 4.329102039337158}
|
| 264 |
+
{"step": 264, "loss": 0.9757599830627441, "grad_norm": 0.6408441662788391, "lr": 6.349294341940593e-06, "epoch": 1, "t": 4.03907036781311}
|
| 265 |
+
{"step": 265, "loss": 1.1450011730194092, "grad_norm": 0.5687059164047241, "lr": 6.092950480945897e-06, "epoch": 1, "t": 4.864096164703369}
|
| 266 |
+
{"step": 266, "loss": 1.0518672466278076, "grad_norm": 0.6086743474006653, "lr": 5.841552533269534e-06, "epoch": 1, "t": 4.132091522216797}
|
| 267 |
+
{"step": 267, "loss": 1.063183307647705, "grad_norm": 0.5224151015281677, "lr": 5.595128817654638e-06, "epoch": 1, "t": 4.723138093948364}
|
| 268 |
+
{"step": 268, "loss": 1.126554012298584, "grad_norm": 0.5347367525100708, "lr": 5.353707092521582e-06, "epoch": 1, "t": 4.741367340087891}
|
| 269 |
+
{"step": 269, "loss": 1.0698590278625488, "grad_norm": 0.5909562110900879, "lr": 5.117314552841052e-06, "epoch": 1, "t": 4.328283309936523}
|
| 270 |
+
{"step": 270, "loss": 0.9937804341316223, "grad_norm": 0.5930314064025879, "lr": 4.885977827070748e-06, "epoch": 1, "t": 4.099947929382324}
|
| 271 |
+
{"step": 271, "loss": 0.9854071140289307, "grad_norm": 0.5892282724380493, "lr": 4.659722974155767e-06, "epoch": 1, "t": 4.106796503067017}
|
| 272 |
+
{"step": 272, "loss": 0.9598199129104614, "grad_norm": 0.5163775682449341, "lr": 4.43857548059321e-06, "epoch": 1, "t": 4.5606396198272705}
|
| 273 |
+
{"step": 273, "loss": 1.0887280702590942, "grad_norm": 0.6039618849754333, "lr": 4.2225602575612755e-06, "epoch": 1, "t": 4.001786231994629}
|
| 274 |
+
{"step": 274, "loss": 0.9749129414558411, "grad_norm": 0.5219699144363403, "lr": 4.011701638113063e-06, "epoch": 1, "t": 5.082299470901489}
|
| 275 |
+
{"step": 275, "loss": 1.0245336294174194, "grad_norm": 0.5433735847473145, "lr": 3.8060233744356633e-06, "epoch": 1, "t": 4.273669719696045}
|
| 276 |
+
{"step": 276, "loss": 1.036287546157837, "grad_norm": 0.6017623543739319, "lr": 3.605548635174533e-06, "epoch": 1, "t": 3.9575648307800293}
|
| 277 |
+
{"step": 277, "loss": 1.08697509765625, "grad_norm": 0.5071364045143127, "lr": 3.410300002823691e-06, "epoch": 1, "t": 5.0819361209869385}
|
| 278 |
+
{"step": 278, "loss": 1.0473910570144653, "grad_norm": 0.573357343673706, "lr": 3.220299471181898e-06, "epoch": 1, "t": 4.301428556442261}
|
| 279 |
+
{"step": 279, "loss": 1.0142626762390137, "grad_norm": 0.47959503531455994, "lr": 3.035568442875136e-06, "epoch": 1, "t": 4.915994167327881}
|
| 280 |
+
{"step": 280, "loss": 1.0749554634094238, "grad_norm": 0.5911064743995667, "lr": 2.85612772694579e-06, "epoch": 1, "t": 4.327515125274658}
|
| 281 |
+
{"step": 281, "loss": 0.943739652633667, "grad_norm": 0.4800060987472534, "lr": 2.6819975365085237e-06, "epoch": 1, "t": 4.712199449539185}
|
| 282 |
+
{"step": 282, "loss": 1.1017746925354004, "grad_norm": 0.6459676027297974, "lr": 2.5131974864734066e-06, "epoch": 1, "t": 3.9813883304595947}
|
| 283 |
+
{"step": 283, "loss": 0.9974814653396606, "grad_norm": 0.6010141968727112, "lr": 2.349746591336405e-06, "epoch": 1, "t": 4.123507499694824}
|
| 284 |
+
{"step": 284, "loss": 1.0443840026855469, "grad_norm": 0.576454222202301, "lr": 2.191663263037458e-06, "epoch": 1, "t": 4.55822229385376}
|
| 285 |
+
{"step": 285, "loss": 1.1475298404693604, "grad_norm": 0.5512248873710632, "lr": 2.0389653088865036e-06, "epoch": 1, "t": 4.435793399810791}
|
| 286 |
+
{"step": 286, "loss": 0.999435544013977, "grad_norm": 0.5110436677932739, "lr": 1.8916699295575324e-06, "epoch": 1, "t": 4.406858682632446}
|
| 287 |
+
{"step": 287, "loss": 1.074988603591919, "grad_norm": 0.5503910779953003, "lr": 1.7497937171510547e-06, "epoch": 1, "t": 4.396028757095337}
|
| 288 |
+
{"step": 288, "loss": 0.9935067892074585, "grad_norm": 0.545520007610321, "lr": 1.6133526533250565e-06, "epoch": 1, "t": 4.584635019302368}
|
| 289 |
+
{"step": 289, "loss": 1.162201166152954, "grad_norm": 0.60584956407547, "lr": 1.4823621074947503e-06, "epoch": 1, "t": 4.111755847930908}
|
| 290 |
+
{"step": 290, "loss": 0.9742871522903442, "grad_norm": 0.484714150428772, "lr": 1.3568368351012717e-06, "epoch": 1, "t": 4.822879791259766}
|
| 291 |
+
{"step": 291, "loss": 1.1879448890686035, "grad_norm": 0.507724940776825, "lr": 1.236790975949592e-06, "epoch": 1, "t": 5.369089603424072}
|
| 292 |
+
{"step": 292, "loss": 1.128878116607666, "grad_norm": 0.549791693687439, "lr": 1.1222380526156928e-06, "epoch": 1, "t": 4.478045225143433}
|
| 293 |
+
{"step": 293, "loss": 1.0033016204833984, "grad_norm": 0.5242128372192383, "lr": 1.0131909689233442e-06, "epoch": 1, "t": 4.921535015106201}
|
| 294 |
+
{"step": 294, "loss": 0.9513107538223267, "grad_norm": 0.6564370393753052, "lr": 9.096620084905472e-07, "epoch": 1, "t": 3.878822088241577}
|
| 295 |
+
{"step": 295, "loss": 0.986213743686676, "grad_norm": 0.5300222635269165, "lr": 8.11662833345822e-07, "epoch": 1, "t": 4.342219352722168}
|
| 296 |
+
{"step": 296, "loss": 1.0823405981063843, "grad_norm": 0.6231135725975037, "lr": 7.192044826145771e-07, "epoch": 1, "t": 3.87117600440979}
|
| 297 |
+
{"step": 297, "loss": 1.0502612590789795, "grad_norm": 0.5496534705162048, "lr": 6.322973712755697e-07, "epoch": 1, "t": 4.422821998596191}
|
| 298 |
+
{"step": 298, "loss": 0.9623367190361023, "grad_norm": 0.567445695400238, "lr": 5.509512889877333e-07, "epoch": 1, "t": 4.143561124801636}
|
| 299 |
+
{"step": 299, "loss": 1.0074646472930908, "grad_norm": 0.5699942111968994, "lr": 4.7517539898741524e-07, "epoch": 1, "t": 4.579758167266846}
|
| 300 |
+
{"step": 300, "loss": 1.0126843452453613, "grad_norm": 0.5032439827919006, "lr": 4.049782370561583e-07, "epoch": 1, "t": 4.58093523979187}
|
| 301 |
+
{"step": 301, "loss": 0.9681507349014282, "grad_norm": 0.5267266035079956, "lr": 3.4036771055923066e-07, "epoch": 1, "t": 4.373915910720825}
|
| 302 |
+
{"step": 302, "loss": 0.9991241693496704, "grad_norm": 0.5616303086280823, "lr": 2.813510975548772e-07, "epoch": 1, "t": 4.088297128677368}
|
| 303 |
+
{"step": 303, "loss": 1.0417941808700562, "grad_norm": 0.5174663066864014, "lr": 2.2793504597447002e-07, "epoch": 1, "t": 4.484217405319214}
|
| 304 |
+
{"step": 304, "loss": 1.0196160078048706, "grad_norm": 0.515720784664154, "lr": 1.8012557287367392e-07, "epoch": 1, "t": 4.566909074783325}
|
| 305 |
+
{"step": 305, "loss": 1.098816156387329, "grad_norm": 0.5197137594223022, "lr": 1.379280637546443e-07, "epoch": 1, "t": 4.583530426025391}
|
| 306 |
+
{"step": 306, "loss": 1.009455680847168, "grad_norm": 0.4882732331752777, "lr": 1.0134727195937333e-07, "epoch": 1, "t": 4.892503499984741}
|
| 307 |
+
{"step": 307, "loss": 0.9840627908706665, "grad_norm": 0.5531430244445801, "lr": 7.038731813426291e-08, "epoch": 1, "t": 4.222560405731201}
|
| 308 |
+
{"step": 308, "loss": 1.1766579151153564, "grad_norm": 0.5558071136474609, "lr": 4.5051689765929214e-08, "epoch": 1, "t": 4.611720085144043}
|
| 309 |
+
{"step": 309, "loss": 1.046766757965088, "grad_norm": 0.5218545794487, "lr": 2.534324078837802e-08, "epoch": 1, "t": 4.333061933517456}
|
| 310 |
+
{"step": 310, "loss": 1.11396324634552, "grad_norm": 0.5571835041046143, "lr": 1.1264191261528557e-08, "epoch": 1, "t": 4.307008981704712}
|
| 311 |
+
{"step": 311, "loss": 1.0265575647354126, "grad_norm": 0.5434891581535339, "lr": 2.8161271211024633e-09, "epoch": 1, "t": 4.516874313354492}
|
| 312 |
+
{"step": 312, "loss": 1.036250114440918, "grad_norm": 0.5302414298057556, "lr": 0.0, "epoch": 1, "t": 4.247893571853638}
|
gemma-3-27b-it/haiku_top-s0/README.md
ADDED
|
@@ -0,0 +1,206 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: google/gemma-3-27b-it
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- base_model:adapter:google/gemma-3-27b-it
|
| 6 |
+
- lora
|
| 7 |
+
- transformers
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# Model Card for Model ID
|
| 11 |
+
|
| 12 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
## Model Details
|
| 17 |
+
|
| 18 |
+
### Model Description
|
| 19 |
+
|
| 20 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
- **Developed by:** [More Information Needed]
|
| 25 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 26 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 27 |
+
- **Model type:** [More Information Needed]
|
| 28 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 29 |
+
- **License:** [More Information Needed]
|
| 30 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 31 |
+
|
| 32 |
+
### Model Sources [optional]
|
| 33 |
+
|
| 34 |
+
<!-- Provide the basic links for the model. -->
|
| 35 |
+
|
| 36 |
+
- **Repository:** [More Information Needed]
|
| 37 |
+
- **Paper [optional]:** [More Information Needed]
|
| 38 |
+
- **Demo [optional]:** [More Information Needed]
|
| 39 |
+
|
| 40 |
+
## Uses
|
| 41 |
+
|
| 42 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 43 |
+
|
| 44 |
+
### Direct Use
|
| 45 |
+
|
| 46 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 47 |
+
|
| 48 |
+
[More Information Needed]
|
| 49 |
+
|
| 50 |
+
### Downstream Use [optional]
|
| 51 |
+
|
| 52 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 53 |
+
|
| 54 |
+
[More Information Needed]
|
| 55 |
+
|
| 56 |
+
### Out-of-Scope Use
|
| 57 |
+
|
| 58 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 59 |
+
|
| 60 |
+
[More Information Needed]
|
| 61 |
+
|
| 62 |
+
## Bias, Risks, and Limitations
|
| 63 |
+
|
| 64 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 65 |
+
|
| 66 |
+
[More Information Needed]
|
| 67 |
+
|
| 68 |
+
### Recommendations
|
| 69 |
+
|
| 70 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 71 |
+
|
| 72 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 73 |
+
|
| 74 |
+
## How to Get Started with the Model
|
| 75 |
+
|
| 76 |
+
Use the code below to get started with the model.
|
| 77 |
+
|
| 78 |
+
[More Information Needed]
|
| 79 |
+
|
| 80 |
+
## Training Details
|
| 81 |
+
|
| 82 |
+
### Training Data
|
| 83 |
+
|
| 84 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 85 |
+
|
| 86 |
+
[More Information Needed]
|
| 87 |
+
|
| 88 |
+
### Training Procedure
|
| 89 |
+
|
| 90 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 91 |
+
|
| 92 |
+
#### Preprocessing [optional]
|
| 93 |
+
|
| 94 |
+
[More Information Needed]
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
#### Training Hyperparameters
|
| 98 |
+
|
| 99 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 100 |
+
|
| 101 |
+
#### Speeds, Sizes, Times [optional]
|
| 102 |
+
|
| 103 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 104 |
+
|
| 105 |
+
[More Information Needed]
|
| 106 |
+
|
| 107 |
+
## Evaluation
|
| 108 |
+
|
| 109 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 110 |
+
|
| 111 |
+
### Testing Data, Factors & Metrics
|
| 112 |
+
|
| 113 |
+
#### Testing Data
|
| 114 |
+
|
| 115 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 116 |
+
|
| 117 |
+
[More Information Needed]
|
| 118 |
+
|
| 119 |
+
#### Factors
|
| 120 |
+
|
| 121 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 122 |
+
|
| 123 |
+
[More Information Needed]
|
| 124 |
+
|
| 125 |
+
#### Metrics
|
| 126 |
+
|
| 127 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 128 |
+
|
| 129 |
+
[More Information Needed]
|
| 130 |
+
|
| 131 |
+
### Results
|
| 132 |
+
|
| 133 |
+
[More Information Needed]
|
| 134 |
+
|
| 135 |
+
#### Summary
|
| 136 |
+
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
## Model Examination [optional]
|
| 140 |
+
|
| 141 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 142 |
+
|
| 143 |
+
[More Information Needed]
|
| 144 |
+
|
| 145 |
+
## Environmental Impact
|
| 146 |
+
|
| 147 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 148 |
+
|
| 149 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 150 |
+
|
| 151 |
+
- **Hardware Type:** [More Information Needed]
|
| 152 |
+
- **Hours used:** [More Information Needed]
|
| 153 |
+
- **Cloud Provider:** [More Information Needed]
|
| 154 |
+
- **Compute Region:** [More Information Needed]
|
| 155 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 156 |
+
|
| 157 |
+
## Technical Specifications [optional]
|
| 158 |
+
|
| 159 |
+
### Model Architecture and Objective
|
| 160 |
+
|
| 161 |
+
[More Information Needed]
|
| 162 |
+
|
| 163 |
+
### Compute Infrastructure
|
| 164 |
+
|
| 165 |
+
[More Information Needed]
|
| 166 |
+
|
| 167 |
+
#### Hardware
|
| 168 |
+
|
| 169 |
+
[More Information Needed]
|
| 170 |
+
|
| 171 |
+
#### Software
|
| 172 |
+
|
| 173 |
+
[More Information Needed]
|
| 174 |
+
|
| 175 |
+
## Citation [optional]
|
| 176 |
+
|
| 177 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 178 |
+
|
| 179 |
+
**BibTeX:**
|
| 180 |
+
|
| 181 |
+
[More Information Needed]
|
| 182 |
+
|
| 183 |
+
**APA:**
|
| 184 |
+
|
| 185 |
+
[More Information Needed]
|
| 186 |
+
|
| 187 |
+
## Glossary [optional]
|
| 188 |
+
|
| 189 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 190 |
+
|
| 191 |
+
[More Information Needed]
|
| 192 |
+
|
| 193 |
+
## More Information [optional]
|
| 194 |
+
|
| 195 |
+
[More Information Needed]
|
| 196 |
+
|
| 197 |
+
## Model Card Authors [optional]
|
| 198 |
+
|
| 199 |
+
[More Information Needed]
|
| 200 |
+
|
| 201 |
+
## Model Card Contact
|
| 202 |
+
|
| 203 |
+
[More Information Needed]
|
| 204 |
+
### Framework versions
|
| 205 |
+
|
| 206 |
+
- PEFT 0.17.1
|
gemma-3-27b-it/haiku_top-s0/adapter_config.json
ADDED
|
@@ -0,0 +1,228 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": {
|
| 4 |
+
"base_model_class": "Gemma3ForConditionalGeneration",
|
| 5 |
+
"parent_library": "transformers.models.gemma3.modeling_gemma3"
|
| 6 |
+
},
|
| 7 |
+
"base_model_name_or_path": "google/gemma-3-27b-it",
|
| 8 |
+
"bias": "none",
|
| 9 |
+
"corda_config": null,
|
| 10 |
+
"eva_config": null,
|
| 11 |
+
"exclude_modules": null,
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": true,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layer_replication": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"layers_to_transform": null,
|
| 18 |
+
"loftq_config": {},
|
| 19 |
+
"lora_alpha": 128,
|
| 20 |
+
"lora_bias": false,
|
| 21 |
+
"lora_dropout": 0.05,
|
| 22 |
+
"megatron_config": null,
|
| 23 |
+
"megatron_core": "megatron.core",
|
| 24 |
+
"modules_to_save": null,
|
| 25 |
+
"peft_type": "LORA",
|
| 26 |
+
"qalora_group_size": 16,
|
| 27 |
+
"r": 64,
|
| 28 |
+
"rank_pattern": {},
|
| 29 |
+
"revision": null,
|
| 30 |
+
"target_modules": [
|
| 31 |
+
"43.self_attn.k_proj",
|
| 32 |
+
"language_model.layers.8.self_attn.v_proj",
|
| 33 |
+
"32.self_attn.q_proj",
|
| 34 |
+
"language_model.layers.20.self_attn.k_proj",
|
| 35 |
+
"language_model.layers.21.self_attn.k_proj",
|
| 36 |
+
"language_model.layers.3.self_attn.q_proj",
|
| 37 |
+
"57.self_attn.q_proj",
|
| 38 |
+
"60.self_attn.v_proj",
|
| 39 |
+
"34.self_attn.k_proj",
|
| 40 |
+
"language_model.layers.10.self_attn.k_proj",
|
| 41 |
+
"54.self_attn.k_proj",
|
| 42 |
+
"48.self_attn.v_proj",
|
| 43 |
+
"37.self_attn.k_proj",
|
| 44 |
+
"48.self_attn.q_proj",
|
| 45 |
+
"41.self_attn.k_proj",
|
| 46 |
+
"51.self_attn.q_proj",
|
| 47 |
+
"language_model.layers.1.self_attn.v_proj",
|
| 48 |
+
"language_model.layers.11.self_attn.q_proj",
|
| 49 |
+
"50.self_attn.v_proj",
|
| 50 |
+
"language_model.layers.18.self_attn.k_proj",
|
| 51 |
+
"language_model.layers.23.self_attn.v_proj",
|
| 52 |
+
"54.self_attn.v_proj",
|
| 53 |
+
"language_model.layers.5.self_attn.k_proj",
|
| 54 |
+
"53.self_attn.v_proj",
|
| 55 |
+
"language_model.layers.9.self_attn.q_proj",
|
| 56 |
+
"language_model.layers.11.self_attn.v_proj",
|
| 57 |
+
"o_proj",
|
| 58 |
+
"34.self_attn.q_proj",
|
| 59 |
+
"language_model.layers.2.self_attn.v_proj",
|
| 60 |
+
"language_model.layers.25.self_attn.v_proj",
|
| 61 |
+
"language_model.layers.2.self_attn.k_proj",
|
| 62 |
+
"language_model.layers.19.self_attn.q_proj",
|
| 63 |
+
"language_model.layers.13.self_attn.q_proj",
|
| 64 |
+
"40.self_attn.v_proj",
|
| 65 |
+
"38.self_attn.v_proj",
|
| 66 |
+
"36.self_attn.q_proj",
|
| 67 |
+
"49.self_attn.k_proj",
|
| 68 |
+
"language_model.layers.12.self_attn.k_proj",
|
| 69 |
+
"61.self_attn.k_proj",
|
| 70 |
+
"29.self_attn.v_proj",
|
| 71 |
+
"38.self_attn.k_proj",
|
| 72 |
+
"language_model.layers.3.self_attn.k_proj",
|
| 73 |
+
"45.self_attn.k_proj",
|
| 74 |
+
"language_model.layers.26.self_attn.k_proj",
|
| 75 |
+
"language_model.layers.22.self_attn.v_proj",
|
| 76 |
+
"39.self_attn.v_proj",
|
| 77 |
+
"30.self_attn.q_proj",
|
| 78 |
+
"42.self_attn.q_proj",
|
| 79 |
+
"30.self_attn.v_proj",
|
| 80 |
+
"58.self_attn.v_proj",
|
| 81 |
+
"54.self_attn.q_proj",
|
| 82 |
+
"45.self_attn.v_proj",
|
| 83 |
+
"60.self_attn.q_proj",
|
| 84 |
+
"language_model.layers.20.self_attn.v_proj",
|
| 85 |
+
"down_proj",
|
| 86 |
+
"language_model.layers.24.self_attn.v_proj",
|
| 87 |
+
"41.self_attn.v_proj",
|
| 88 |
+
"language_model.layers.15.self_attn.k_proj",
|
| 89 |
+
"language_model.layers.0.self_attn.q_proj",
|
| 90 |
+
"46.self_attn.q_proj",
|
| 91 |
+
"27.self_attn.q_proj",
|
| 92 |
+
"language_model.layers.26.self_attn.v_proj",
|
| 93 |
+
"51.self_attn.v_proj",
|
| 94 |
+
"language_model.layers.5.self_attn.q_proj",
|
| 95 |
+
"59.self_attn.q_proj",
|
| 96 |
+
"44.self_attn.v_proj",
|
| 97 |
+
"47.self_attn.q_proj",
|
| 98 |
+
"language_model.layers.8.self_attn.k_proj",
|
| 99 |
+
"60.self_attn.k_proj",
|
| 100 |
+
"language_model.layers.8.self_attn.q_proj",
|
| 101 |
+
"language_model.layers.18.self_attn.v_proj",
|
| 102 |
+
"language_model.layers.10.self_attn.v_proj",
|
| 103 |
+
"language_model.layers.16.self_attn.k_proj",
|
| 104 |
+
"39.self_attn.k_proj",
|
| 105 |
+
"language_model.layers.4.self_attn.k_proj",
|
| 106 |
+
"28.self_attn.v_proj",
|
| 107 |
+
"38.self_attn.q_proj",
|
| 108 |
+
"56.self_attn.q_proj",
|
| 109 |
+
"32.self_attn.v_proj",
|
| 110 |
+
"34.self_attn.v_proj",
|
| 111 |
+
"56.self_attn.v_proj",
|
| 112 |
+
"up_proj",
|
| 113 |
+
"55.self_attn.v_proj",
|
| 114 |
+
"46.self_attn.k_proj",
|
| 115 |
+
"33.self_attn.k_proj",
|
| 116 |
+
"language_model.layers.20.self_attn.q_proj",
|
| 117 |
+
"language_model.layers.4.self_attn.q_proj",
|
| 118 |
+
"53.self_attn.q_proj",
|
| 119 |
+
"45.self_attn.q_proj",
|
| 120 |
+
"language_model.layers.0.self_attn.k_proj",
|
| 121 |
+
"40.self_attn.k_proj",
|
| 122 |
+
"language_model.layers.22.self_attn.k_proj",
|
| 123 |
+
"language_model.layers.16.self_attn.q_proj",
|
| 124 |
+
"44.self_attn.k_proj",
|
| 125 |
+
"language_model.layers.18.self_attn.q_proj",
|
| 126 |
+
"35.self_attn.q_proj",
|
| 127 |
+
"49.self_attn.q_proj",
|
| 128 |
+
"35.self_attn.v_proj",
|
| 129 |
+
"52.self_attn.k_proj",
|
| 130 |
+
"29.self_attn.q_proj",
|
| 131 |
+
"58.self_attn.k_proj",
|
| 132 |
+
"language_model.layers.14.self_attn.k_proj",
|
| 133 |
+
"language_model.layers.21.self_attn.q_proj",
|
| 134 |
+
"56.self_attn.k_proj",
|
| 135 |
+
"47.self_attn.v_proj",
|
| 136 |
+
"55.self_attn.k_proj",
|
| 137 |
+
"59.self_attn.v_proj",
|
| 138 |
+
"31.self_attn.k_proj",
|
| 139 |
+
"language_model.layers.10.self_attn.q_proj",
|
| 140 |
+
"51.self_attn.k_proj",
|
| 141 |
+
"language_model.layers.6.self_attn.v_proj",
|
| 142 |
+
"33.self_attn.v_proj",
|
| 143 |
+
"language_model.layers.19.self_attn.v_proj",
|
| 144 |
+
"31.self_attn.v_proj",
|
| 145 |
+
"language_model.layers.7.self_attn.v_proj",
|
| 146 |
+
"39.self_attn.q_proj",
|
| 147 |
+
"29.self_attn.k_proj",
|
| 148 |
+
"57.self_attn.k_proj",
|
| 149 |
+
"language_model.layers.25.self_attn.q_proj",
|
| 150 |
+
"language_model.layers.15.self_attn.v_proj",
|
| 151 |
+
"language_model.layers.6.self_attn.q_proj",
|
| 152 |
+
"language_model.layers.1.self_attn.k_proj",
|
| 153 |
+
"31.self_attn.q_proj",
|
| 154 |
+
"language_model.layers.17.self_attn.v_proj",
|
| 155 |
+
"55.self_attn.q_proj",
|
| 156 |
+
"language_model.layers.24.self_attn.q_proj",
|
| 157 |
+
"language_model.layers.7.self_attn.k_proj",
|
| 158 |
+
"50.self_attn.k_proj",
|
| 159 |
+
"language_model.layers.0.self_attn.v_proj",
|
| 160 |
+
"language_model.layers.5.self_attn.v_proj",
|
| 161 |
+
"42.self_attn.v_proj",
|
| 162 |
+
"27.self_attn.v_proj",
|
| 163 |
+
"48.self_attn.k_proj",
|
| 164 |
+
"40.self_attn.q_proj",
|
| 165 |
+
"language_model.layers.14.self_attn.v_proj",
|
| 166 |
+
"language_model.layers.12.self_attn.v_proj",
|
| 167 |
+
"58.self_attn.q_proj",
|
| 168 |
+
"language_model.layers.9.self_attn.v_proj",
|
| 169 |
+
"36.self_attn.v_proj",
|
| 170 |
+
"language_model.layers.17.self_attn.q_proj",
|
| 171 |
+
"44.self_attn.q_proj",
|
| 172 |
+
"language_model.layers.9.self_attn.k_proj",
|
| 173 |
+
"language_model.layers.11.self_attn.k_proj",
|
| 174 |
+
"language_model.layers.15.self_attn.q_proj",
|
| 175 |
+
"53.self_attn.k_proj",
|
| 176 |
+
"language_model.layers.13.self_attn.k_proj",
|
| 177 |
+
"language_model.layers.23.self_attn.q_proj",
|
| 178 |
+
"language_model.layers.4.self_attn.v_proj",
|
| 179 |
+
"30.self_attn.k_proj",
|
| 180 |
+
"gate_proj",
|
| 181 |
+
"language_model.layers.6.self_attn.k_proj",
|
| 182 |
+
"52.self_attn.v_proj",
|
| 183 |
+
"language_model.layers.13.self_attn.v_proj",
|
| 184 |
+
"33.self_attn.q_proj",
|
| 185 |
+
"37.self_attn.v_proj",
|
| 186 |
+
"language_model.layers.26.self_attn.q_proj",
|
| 187 |
+
"59.self_attn.k_proj",
|
| 188 |
+
"language_model.layers.25.self_attn.k_proj",
|
| 189 |
+
"language_model.layers.14.self_attn.q_proj",
|
| 190 |
+
"language_model.layers.16.self_attn.v_proj",
|
| 191 |
+
"43.self_attn.q_proj",
|
| 192 |
+
"language_model.layers.24.self_attn.k_proj",
|
| 193 |
+
"language_model.layers.12.self_attn.q_proj",
|
| 194 |
+
"61.self_attn.q_proj",
|
| 195 |
+
"43.self_attn.v_proj",
|
| 196 |
+
"42.self_attn.k_proj",
|
| 197 |
+
"61.self_attn.v_proj",
|
| 198 |
+
"28.self_attn.k_proj",
|
| 199 |
+
"28.self_attn.q_proj",
|
| 200 |
+
"36.self_attn.k_proj",
|
| 201 |
+
"language_model.layers.22.self_attn.q_proj",
|
| 202 |
+
"language_model.layers.1.self_attn.q_proj",
|
| 203 |
+
"47.self_attn.k_proj",
|
| 204 |
+
"52.self_attn.q_proj",
|
| 205 |
+
"37.self_attn.q_proj",
|
| 206 |
+
"57.self_attn.v_proj",
|
| 207 |
+
"27.self_attn.k_proj",
|
| 208 |
+
"language_model.layers.3.self_attn.v_proj",
|
| 209 |
+
"41.self_attn.q_proj",
|
| 210 |
+
"50.self_attn.q_proj",
|
| 211 |
+
"35.self_attn.k_proj",
|
| 212 |
+
"49.self_attn.v_proj",
|
| 213 |
+
"46.self_attn.v_proj",
|
| 214 |
+
"language_model.layers.17.self_attn.k_proj",
|
| 215 |
+
"language_model.layers.23.self_attn.k_proj",
|
| 216 |
+
"language_model.layers.2.self_attn.q_proj",
|
| 217 |
+
"language_model.layers.7.self_attn.q_proj",
|
| 218 |
+
"language_model.layers.21.self_attn.v_proj",
|
| 219 |
+
"32.self_attn.k_proj",
|
| 220 |
+
"language_model.layers.19.self_attn.k_proj"
|
| 221 |
+
],
|
| 222 |
+
"target_parameters": null,
|
| 223 |
+
"task_type": null,
|
| 224 |
+
"trainable_token_indices": null,
|
| 225 |
+
"use_dora": false,
|
| 226 |
+
"use_qalora": false,
|
| 227 |
+
"use_rslora": false
|
| 228 |
+
}
|
gemma-3-27b-it/haiku_top-s0/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5423dfd9dede6b805cdd3362607ba0ad1c5c2e7cb7d9713e7be3481e0da8100a
|
| 3 |
+
size 1816396048
|
gemma-3-27b-it/haiku_top-s0/config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "google/gemma-3-27b-it",
|
| 3 |
+
"data": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma/arms/TASKW-haiku_top-gemma.jsonl",
|
| 4 |
+
"run": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma/runs/haiku_top-s0",
|
| 5 |
+
"steps": 0,
|
| 6 |
+
"batch": 16,
|
| 7 |
+
"micro": 1,
|
| 8 |
+
"seq_len": 4096,
|
| 9 |
+
"lr": 0.0001,
|
| 10 |
+
"warmup": 10,
|
| 11 |
+
"lora_rank": 64,
|
| 12 |
+
"lora_targets": "all-linear",
|
| 13 |
+
"lora_targets_file": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/code/gemma3_lora_targets.json",
|
| 14 |
+
"seed": 0,
|
| 15 |
+
"bos": true,
|
| 16 |
+
"no_merge": false,
|
| 17 |
+
"save_at": "",
|
| 18 |
+
"replay": "",
|
| 19 |
+
"replay_frac": 0.0,
|
| 20 |
+
"constant_lr": false,
|
| 21 |
+
"schedule": "hf-cosine",
|
| 22 |
+
"min_lr_frac": 0.0,
|
| 23 |
+
"warmup_ratio": 0.05,
|
| 24 |
+
"lora_dropout": 0.05,
|
| 25 |
+
"no_pack": true,
|
| 26 |
+
"chat_sft": true,
|
| 27 |
+
"data_frac": 1.0,
|
| 28 |
+
"base_adapter": "",
|
| 29 |
+
"ckpt_every": 0,
|
| 30 |
+
"kill_after": 0,
|
| 31 |
+
"method": "doc_lm_lora",
|
| 32 |
+
"world": 2
|
| 33 |
+
}
|
gemma-3-27b-it/haiku_top-s0/log.jsonl
ADDED
|
@@ -0,0 +1,312 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 1, "loss": 2.3165102005004883, "grad_norm": 4.408999443054199, "lr": 1.25e-05, "epoch": 1, "t": 5.280462741851807}
|
| 2 |
+
{"step": 2, "loss": 2.336566925048828, "grad_norm": 4.590346336364746, "lr": 1.8750000000000002e-05, "epoch": 1, "t": 4.5219056606292725}
|
| 3 |
+
{"step": 3, "loss": 2.206780433654785, "grad_norm": 3.9203405380249023, "lr": 2.5e-05, "epoch": 1, "t": 4.73903489112854}
|
| 4 |
+
{"step": 4, "loss": 1.8827847242355347, "grad_norm": 3.0371716022491455, "lr": 3.125e-05, "epoch": 1, "t": 4.430527687072754}
|
| 5 |
+
{"step": 5, "loss": 1.924107551574707, "grad_norm": 2.5753631591796875, "lr": 3.7500000000000003e-05, "epoch": 1, "t": 4.00973653793335}
|
| 6 |
+
{"step": 6, "loss": 1.5902197360992432, "grad_norm": 1.8982183933258057, "lr": 4.375e-05, "epoch": 1, "t": 4.1120805740356445}
|
| 7 |
+
{"step": 7, "loss": 1.436042070388794, "grad_norm": 0.8262069821357727, "lr": 5e-05, "epoch": 1, "t": 4.713772296905518}
|
| 8 |
+
{"step": 8, "loss": 1.4866223335266113, "grad_norm": 0.7413012981414795, "lr": 5.6250000000000005e-05, "epoch": 1, "t": 4.541523694992065}
|
| 9 |
+
{"step": 9, "loss": 1.4765052795410156, "grad_norm": 0.7449035048484802, "lr": 6.25e-05, "epoch": 1, "t": 4.6371705532073975}
|
| 10 |
+
{"step": 10, "loss": 1.4025652408599854, "grad_norm": 0.7779823541641235, "lr": 6.875e-05, "epoch": 1, "t": 4.777430295944214}
|
| 11 |
+
{"step": 11, "loss": 1.3516591787338257, "grad_norm": 0.8012948632240295, "lr": 7.500000000000001e-05, "epoch": 1, "t": 4.253003358840942}
|
| 12 |
+
{"step": 12, "loss": 1.470594882965088, "grad_norm": 0.7655975222587585, "lr": 8.125000000000001e-05, "epoch": 1, "t": 5.353244066238403}
|
| 13 |
+
{"step": 13, "loss": 1.2650811672210693, "grad_norm": 0.6489203572273254, "lr": 8.75e-05, "epoch": 1, "t": 4.3800764083862305}
|
| 14 |
+
{"step": 14, "loss": 1.284051537513733, "grad_norm": 0.6479799151420593, "lr": 9.375e-05, "epoch": 1, "t": 4.319174289703369}
|
| 15 |
+
{"step": 15, "loss": 1.38352370262146, "grad_norm": 0.5979617238044739, "lr": 0.0001, "epoch": 1, "t": 4.757409334182739}
|
| 16 |
+
{"step": 16, "loss": 1.3490856885910034, "grad_norm": 0.63264000415802, "lr": 0.0001, "epoch": 1, "t": 4.596350193023682}
|
| 17 |
+
{"step": 17, "loss": 1.4035805463790894, "grad_norm": 0.6797512173652649, "lr": 9.99971838728789e-05, "epoch": 1, "t": 4.694079875946045}
|
| 18 |
+
{"step": 18, "loss": 1.3759143352508545, "grad_norm": 0.6469725370407104, "lr": 9.998873580873848e-05, "epoch": 1, "t": 4.708922863006592}
|
| 19 |
+
{"step": 19, "loss": 1.2821584939956665, "grad_norm": 0.5781360864639282, "lr": 9.997465675921163e-05, "epoch": 1, "t": 4.854948997497559}
|
| 20 |
+
{"step": 20, "loss": 1.2673040628433228, "grad_norm": 0.536912202835083, "lr": 9.995494831023409e-05, "epoch": 1, "t": 4.835841417312622}
|
| 21 |
+
{"step": 21, "loss": 1.2788596153259277, "grad_norm": 0.5443888902664185, "lr": 9.992961268186573e-05, "epoch": 1, "t": 4.4762513637542725}
|
| 22 |
+
{"step": 22, "loss": 1.2426241636276245, "grad_norm": 0.5355674624443054, "lr": 9.989865272804063e-05, "epoch": 1, "t": 5.172754764556885}
|
| 23 |
+
{"step": 23, "loss": 1.2968319654464722, "grad_norm": 0.5094603300094604, "lr": 9.986207193624536e-05, "epoch": 1, "t": 5.170877933502197}
|
| 24 |
+
{"step": 24, "loss": 1.220749855041504, "grad_norm": 0.5117735862731934, "lr": 9.981987442712633e-05, "epoch": 1, "t": 4.705889940261841}
|
| 25 |
+
{"step": 25, "loss": 1.2357466220855713, "grad_norm": 0.5151938796043396, "lr": 9.977206495402554e-05, "epoch": 1, "t": 4.686975002288818}
|
| 26 |
+
{"step": 26, "loss": 1.225297451019287, "grad_norm": 0.5193535685539246, "lr": 9.971864890244513e-05, "epoch": 1, "t": 4.4568774700164795}
|
| 27 |
+
{"step": 27, "loss": 1.1733131408691406, "grad_norm": 0.5163609981536865, "lr": 9.965963228944078e-05, "epoch": 1, "t": 4.786885023117065}
|
| 28 |
+
{"step": 28, "loss": 1.1244175434112549, "grad_norm": 0.5233328342437744, "lr": 9.959502176294383e-05, "epoch": 1, "t": 5.150769233703613}
|
| 29 |
+
{"step": 29, "loss": 1.143699288368225, "grad_norm": 0.5393513441085815, "lr": 9.95248246010126e-05, "epoch": 1, "t": 4.708935260772705}
|
| 30 |
+
{"step": 30, "loss": 1.2240262031555176, "grad_norm": 0.5021150708198547, "lr": 9.944904871101228e-05, "epoch": 1, "t": 5.6406919956207275}
|
| 31 |
+
{"step": 31, "loss": 1.2625977993011475, "grad_norm": 0.5807600617408752, "lr": 9.936770262872443e-05, "epoch": 1, "t": 4.4734556674957275}
|
| 32 |
+
{"step": 32, "loss": 1.2335271835327148, "grad_norm": 0.5684121251106262, "lr": 9.928079551738543e-05, "epoch": 1, "t": 4.456062078475952}
|
| 33 |
+
{"step": 33, "loss": 1.1643261909484863, "grad_norm": 0.5546866655349731, "lr": 9.918833716665419e-05, "epoch": 1, "t": 4.302427768707275}
|
| 34 |
+
{"step": 34, "loss": 1.2406728267669678, "grad_norm": 0.5687874555587769, "lr": 9.909033799150946e-05, "epoch": 1, "t": 4.514441013336182}
|
| 35 |
+
{"step": 35, "loss": 1.1501606702804565, "grad_norm": 0.5952447652816772, "lr": 9.898680903107666e-05, "epoch": 1, "t": 5.107458114624023}
|
| 36 |
+
{"step": 36, "loss": 1.115567922592163, "grad_norm": 0.5387961268424988, "lr": 9.887776194738432e-05, "epoch": 1, "t": 4.893448829650879}
|
| 37 |
+
{"step": 37, "loss": 1.002673864364624, "grad_norm": 0.49590620398521423, "lr": 9.876320902405042e-05, "epoch": 1, "t": 4.81894063949585}
|
| 38 |
+
{"step": 38, "loss": 1.29767906665802, "grad_norm": 0.5549535751342773, "lr": 9.864316316489873e-05, "epoch": 1, "t": 5.790158033370972}
|
| 39 |
+
{"step": 39, "loss": 1.118459701538086, "grad_norm": 0.5180652737617493, "lr": 9.851763789250525e-05, "epoch": 1, "t": 5.822343587875366}
|
| 40 |
+
{"step": 40, "loss": 1.2622766494750977, "grad_norm": 0.5744220614433289, "lr": 9.838664734667495e-05, "epoch": 1, "t": 5.820775270462036}
|
| 41 |
+
{"step": 41, "loss": 1.2271785736083984, "grad_norm": 0.544512152671814, "lr": 9.825020628284896e-05, "epoch": 1, "t": 5.9317827224731445}
|
| 42 |
+
{"step": 42, "loss": 1.167788028717041, "grad_norm": 0.5220765471458435, "lr": 9.810833007044247e-05, "epoch": 1, "t": 6.025085926055908}
|
| 43 |
+
{"step": 43, "loss": 1.0724492073059082, "grad_norm": 0.48651739954948425, "lr": 9.796103469111351e-05, "epoch": 1, "t": 5.295976877212524}
|
| 44 |
+
{"step": 44, "loss": 1.0941650867462158, "grad_norm": 0.5707746744155884, "lr": 9.780833673696254e-05, "epoch": 1, "t": 4.358911514282227}
|
| 45 |
+
{"step": 45, "loss": 1.0769819021224976, "grad_norm": 0.5811222195625305, "lr": 9.76502534086636e-05, "epoch": 1, "t": 4.390970230102539}
|
| 46 |
+
{"step": 46, "loss": 1.0222127437591553, "grad_norm": 0.5258973240852356, "lr": 9.74868025135266e-05, "epoch": 1, "t": 4.817179441452026}
|
| 47 |
+
{"step": 47, "loss": 1.3102638721466064, "grad_norm": 0.5098415017127991, "lr": 9.731800246349148e-05, "epoch": 1, "t": 5.175843000411987}
|
| 48 |
+
{"step": 48, "loss": 1.2075951099395752, "grad_norm": 0.5125856995582581, "lr": 9.714387227305422e-05, "epoch": 1, "t": 5.421111106872559}
|
| 49 |
+
{"step": 49, "loss": 1.1449198722839355, "grad_norm": 0.5123289227485657, "lr": 9.696443155712486e-05, "epoch": 1, "t": 5.5788233280181885}
|
| 50 |
+
{"step": 50, "loss": 1.1364178657531738, "grad_norm": 0.5330806374549866, "lr": 9.67797005288181e-05, "epoch": 1, "t": 5.3625054359436035}
|
| 51 |
+
{"step": 51, "loss": 1.1169184446334839, "grad_norm": 0.6335769295692444, "lr": 9.65896999971763e-05, "epoch": 1, "t": 5.759647846221924}
|
| 52 |
+
{"step": 52, "loss": 1.1893417835235596, "grad_norm": 0.5339292287826538, "lr": 9.639445136482548e-05, "epoch": 1, "t": 5.879992485046387}
|
| 53 |
+
{"step": 53, "loss": 1.1272997856140137, "grad_norm": 0.6444324254989624, "lr": 9.619397662556435e-05, "epoch": 1, "t": 5.895978212356567}
|
| 54 |
+
{"step": 54, "loss": 1.1151115894317627, "grad_norm": 0.5933447480201721, "lr": 9.598829836188694e-05, "epoch": 1, "t": 5.9767396450042725}
|
| 55 |
+
{"step": 55, "loss": 1.1766557693481445, "grad_norm": 0.5371019244194031, "lr": 9.577743974243874e-05, "epoch": 1, "t": 6.158986806869507}
|
| 56 |
+
{"step": 56, "loss": 1.1404342651367188, "grad_norm": 0.5204576849937439, "lr": 9.55614245194068e-05, "epoch": 1, "t": 6.1354851722717285}
|
| 57 |
+
{"step": 57, "loss": 1.2516920566558838, "grad_norm": 0.5574812293052673, "lr": 9.534027702584425e-05, "epoch": 1, "t": 6.045910358428955}
|
| 58 |
+
{"step": 58, "loss": 1.0862247943878174, "grad_norm": 0.5740494132041931, "lr": 9.511402217292926e-05, "epoch": 1, "t": 5.915088891983032}
|
| 59 |
+
{"step": 59, "loss": 1.1865355968475342, "grad_norm": 0.5267467498779297, "lr": 9.488268544715896e-05, "epoch": 1, "t": 6.236257791519165}
|
| 60 |
+
{"step": 60, "loss": 1.1413038969039917, "grad_norm": 0.5235229730606079, "lr": 9.464629290747842e-05, "epoch": 1, "t": 5.819396257400513}
|
| 61 |
+
{"step": 61, "loss": 0.9996364712715149, "grad_norm": 0.529492199420929, "lr": 9.440487118234535e-05, "epoch": 1, "t": 5.958328008651733}
|
| 62 |
+
{"step": 62, "loss": 1.0883084535598755, "grad_norm": 0.5600897073745728, "lr": 9.415844746673047e-05, "epoch": 1, "t": 5.942508935928345}
|
| 63 |
+
{"step": 63, "loss": 1.113850474357605, "grad_norm": 0.5150932669639587, "lr": 9.390704951905411e-05, "epoch": 1, "t": 6.093647480010986}
|
| 64 |
+
{"step": 64, "loss": 1.1780354976654053, "grad_norm": 0.5905842185020447, "lr": 9.365070565805941e-05, "epoch": 1, "t": 5.866363048553467}
|
| 65 |
+
{"step": 65, "loss": 1.2376737594604492, "grad_norm": 0.5111215114593506, "lr": 9.338944475962237e-05, "epoch": 1, "t": 6.301398038864136}
|
| 66 |
+
{"step": 66, "loss": 1.0892138481140137, "grad_norm": 0.5049899816513062, "lr": 9.312329625349902e-05, "epoch": 1, "t": 5.9231343269348145}
|
| 67 |
+
{"step": 67, "loss": 1.120809555053711, "grad_norm": 0.5145349502563477, "lr": 9.285229012001047e-05, "epoch": 1, "t": 5.963319540023804}
|
| 68 |
+
{"step": 68, "loss": 1.0082751512527466, "grad_norm": 0.6743009686470032, "lr": 9.257645688666556e-05, "epoch": 1, "t": 5.9756247997283936}
|
| 69 |
+
{"step": 69, "loss": 1.1106818914413452, "grad_norm": 0.5815789103507996, "lr": 9.22958276247223e-05, "epoch": 1, "t": 5.945407390594482}
|
| 70 |
+
{"step": 70, "loss": 1.1567025184631348, "grad_norm": 0.49667027592658997, "lr": 9.201043394568773e-05, "epoch": 1, "t": 6.419935464859009}
|
| 71 |
+
{"step": 71, "loss": 1.2616078853607178, "grad_norm": 0.548518717288971, "lr": 9.172030799775699e-05, "epoch": 1, "t": 6.252639055252075}
|
| 72 |
+
{"step": 72, "loss": 1.0600035190582275, "grad_norm": 0.5150569081306458, "lr": 9.142548246219212e-05, "epoch": 1, "t": 5.968430995941162}
|
| 73 |
+
{"step": 73, "loss": 0.9821887612342834, "grad_norm": 0.5481364727020264, "lr": 9.112599054964057e-05, "epoch": 1, "t": 5.89496111869812}
|
| 74 |
+
{"step": 74, "loss": 1.2077808380126953, "grad_norm": 0.5297751426696777, "lr": 9.082186599639428e-05, "epoch": 1, "t": 6.027462959289551}
|
| 75 |
+
{"step": 75, "loss": 1.1566251516342163, "grad_norm": 0.523955225944519, "lr": 9.051314306058933e-05, "epoch": 1, "t": 6.080850124359131}
|
| 76 |
+
{"step": 76, "loss": 1.1023098230361938, "grad_norm": 0.49539417028427124, "lr": 9.019985651834703e-05, "epoch": 1, "t": 6.159868240356445}
|
| 77 |
+
{"step": 77, "loss": 1.0896000862121582, "grad_norm": 0.518477201461792, "lr": 8.988204165985649e-05, "epoch": 1, "t": 5.896232843399048}
|
| 78 |
+
{"step": 78, "loss": 1.1309046745300293, "grad_norm": 0.5526289939880371, "lr": 8.955973428539944e-05, "epoch": 1, "t": 5.843094110488892}
|
| 79 |
+
{"step": 79, "loss": 1.0300159454345703, "grad_norm": 0.504365086555481, "lr": 8.923297070131737e-05, "epoch": 1, "t": 5.850951671600342}
|
| 80 |
+
{"step": 80, "loss": 1.245361089706421, "grad_norm": 0.8191688656806946, "lr": 8.890178771592199e-05, "epoch": 1, "t": 6.2001423835754395}
|
| 81 |
+
{"step": 81, "loss": 1.039421558380127, "grad_norm": 0.5981826186180115, "lr": 8.856622263534875e-05, "epoch": 1, "t": 5.884080171585083}
|
| 82 |
+
{"step": 82, "loss": 1.092210292816162, "grad_norm": 0.5935723781585693, "lr": 8.822631325935463e-05, "epoch": 1, "t": 5.816681146621704}
|
| 83 |
+
{"step": 83, "loss": 1.0377435684204102, "grad_norm": 0.5498448014259338, "lr": 8.788209787706015e-05, "epoch": 1, "t": 5.803557872772217}
|
| 84 |
+
{"step": 84, "loss": 1.0781831741333008, "grad_norm": 0.5144972801208496, "lr": 8.753361526263621e-05, "epoch": 1, "t": 5.8382182121276855}
|
| 85 |
+
{"step": 85, "loss": 1.1457104682922363, "grad_norm": 0.5221476554870605, "lr": 8.718090467093654e-05, "epoch": 1, "t": 6.018940687179565}
|
| 86 |
+
{"step": 86, "loss": 1.0541032552719116, "grad_norm": 0.5313003659248352, "lr": 8.682400583307562e-05, "epoch": 1, "t": 5.9045398235321045}
|
| 87 |
+
{"step": 87, "loss": 1.0966575145721436, "grad_norm": 0.8335580825805664, "lr": 8.646295895195333e-05, "epoch": 1, "t": 5.809002161026001}
|
| 88 |
+
{"step": 88, "loss": 1.1195096969604492, "grad_norm": 0.529839277267456, "lr": 8.609780469772623e-05, "epoch": 1, "t": 5.913843631744385}
|
| 89 |
+
{"step": 89, "loss": 1.0631704330444336, "grad_norm": 0.5742178559303284, "lr": 8.572858420322627e-05, "epoch": 1, "t": 4.8021955490112305}
|
| 90 |
+
{"step": 90, "loss": 1.1303542852401733, "grad_norm": 0.5403726100921631, "lr": 8.535533905932738e-05, "epoch": 1, "t": 5.146021366119385}
|
| 91 |
+
{"step": 91, "loss": 1.029839277267456, "grad_norm": 0.4675491750240326, "lr": 8.497811131026046e-05, "epoch": 1, "t": 5.039720058441162}
|
| 92 |
+
{"step": 92, "loss": 1.0687185525894165, "grad_norm": 0.5150817036628723, "lr": 8.459694344887732e-05, "epoch": 1, "t": 5.127553462982178}
|
| 93 |
+
{"step": 93, "loss": 1.0635809898376465, "grad_norm": 0.5708933472633362, "lr": 8.421187841186402e-05, "epoch": 1, "t": 4.1637444496154785}
|
| 94 |
+
{"step": 94, "loss": 1.0670030117034912, "grad_norm": 0.5571341514587402, "lr": 8.382295957490436e-05, "epoch": 1, "t": 5.213702917098999}
|
| 95 |
+
{"step": 95, "loss": 1.166431188583374, "grad_norm": 0.5358306169509888, "lr": 8.343023074779368e-05, "epoch": 1, "t": 4.473605155944824}
|
| 96 |
+
{"step": 96, "loss": 1.0168753862380981, "grad_norm": 0.5470162034034729, "lr": 8.303373616950408e-05, "epoch": 1, "t": 4.4981913566589355}
|
| 97 |
+
{"step": 97, "loss": 1.1562048196792603, "grad_norm": 0.539106011390686, "lr": 8.263352050320094e-05, "epoch": 1, "t": 5.114293336868286}
|
| 98 |
+
{"step": 98, "loss": 1.0642127990722656, "grad_norm": 0.49973928928375244, "lr": 8.222962883121196e-05, "epoch": 1, "t": 5.822341442108154}
|
| 99 |
+
{"step": 99, "loss": 1.0666629076004028, "grad_norm": 0.584915280342102, "lr": 8.182210664994878e-05, "epoch": 1, "t": 5.123532056808472}
|
| 100 |
+
{"step": 100, "loss": 0.8964484930038452, "grad_norm": 0.5403911471366882, "lr": 8.141099986478212e-05, "epoch": 1, "t": 4.925328254699707}
|
| 101 |
+
{"step": 101, "loss": 1.088526725769043, "grad_norm": 0.4972061216831207, "lr": 8.099635478487064e-05, "epoch": 1, "t": 4.823850870132446}
|
| 102 |
+
{"step": 102, "loss": 1.1033992767333984, "grad_norm": 0.5467894077301025, "lr": 8.057821811794458e-05, "epoch": 1, "t": 4.284470558166504}
|
| 103 |
+
{"step": 103, "loss": 1.0869007110595703, "grad_norm": 0.5603974461555481, "lr": 8.015663696504422e-05, "epoch": 1, "t": 4.7065417766571045}
|
| 104 |
+
{"step": 104, "loss": 1.010480284690857, "grad_norm": 0.5039533376693726, "lr": 7.973165881521434e-05, "epoch": 1, "t": 4.456177234649658}
|
| 105 |
+
{"step": 105, "loss": 1.1642184257507324, "grad_norm": 0.5268582701683044, "lr": 7.930333154015466e-05, "epoch": 1, "t": 4.695672512054443}
|
| 106 |
+
{"step": 106, "loss": 1.1423639059066772, "grad_norm": 0.5723770260810852, "lr": 7.88717033888274e-05, "epoch": 1, "t": 4.574386358261108}
|
| 107 |
+
{"step": 107, "loss": 1.029895305633545, "grad_norm": 0.5344498753547668, "lr": 7.843682298202235e-05, "epoch": 1, "t": 4.394898414611816}
|
| 108 |
+
{"step": 108, "loss": 1.1407885551452637, "grad_norm": 0.5181390643119812, "lr": 7.799873930687978e-05, "epoch": 1, "t": 4.726134300231934}
|
| 109 |
+
{"step": 109, "loss": 1.080150842666626, "grad_norm": 0.528947114944458, "lr": 7.755750171137246e-05, "epoch": 1, "t": 4.560563564300537}
|
| 110 |
+
{"step": 110, "loss": 1.1265102624893188, "grad_norm": 0.5478967428207397, "lr": 7.711315989874677e-05, "epoch": 1, "t": 4.508591890335083}
|
| 111 |
+
{"step": 111, "loss": 1.089923620223999, "grad_norm": 0.5973758697509766, "lr": 7.666576392192389e-05, "epoch": 1, "t": 4.075223922729492}
|
| 112 |
+
{"step": 112, "loss": 1.1726192235946655, "grad_norm": 0.5653948187828064, "lr": 7.621536417786159e-05, "epoch": 1, "t": 4.137712717056274}
|
| 113 |
+
{"step": 113, "loss": 1.1260628700256348, "grad_norm": 0.528526246547699, "lr": 7.576201140187727e-05, "epoch": 1, "t": 5.274724960327148}
|
| 114 |
+
{"step": 114, "loss": 1.122440218925476, "grad_norm": 0.5097289681434631, "lr": 7.530575666193283e-05, "epoch": 1, "t": 4.944204092025757}
|
| 115 |
+
{"step": 115, "loss": 1.1147640943527222, "grad_norm": 0.5409600138664246, "lr": 7.484665135288213e-05, "epoch": 1, "t": 4.766234397888184}
|
| 116 |
+
{"step": 116, "loss": 1.1317052841186523, "grad_norm": 0.5394009947776794, "lr": 7.438474719068173e-05, "epoch": 1, "t": 5.168792009353638}
|
| 117 |
+
{"step": 117, "loss": 1.1236785650253296, "grad_norm": 0.5077928304672241, "lr": 7.392009620656513e-05, "epoch": 1, "t": 5.529633522033691}
|
| 118 |
+
{"step": 118, "loss": 1.07792329788208, "grad_norm": 0.522649347782135, "lr": 7.345275074118185e-05, "epoch": 1, "t": 5.236745119094849}
|
| 119 |
+
{"step": 119, "loss": 1.025869607925415, "grad_norm": 0.5037496089935303, "lr": 7.298276343870151e-05, "epoch": 1, "t": 5.143819808959961}
|
| 120 |
+
{"step": 120, "loss": 1.0352202653884888, "grad_norm": 0.48912253975868225, "lr": 7.251018724088367e-05, "epoch": 1, "t": 4.872535228729248}
|
| 121 |
+
{"step": 121, "loss": 1.1937408447265625, "grad_norm": 0.5337826609611511, "lr": 7.203507538111423e-05, "epoch": 1, "t": 4.5598907470703125}
|
| 122 |
+
{"step": 122, "loss": 1.0413992404937744, "grad_norm": 0.45708900690078735, "lr": 7.155748137840892e-05, "epoch": 1, "t": 4.897008657455444}
|
| 123 |
+
{"step": 123, "loss": 1.049370288848877, "grad_norm": 0.5275327563285828, "lr": 7.107745903138472e-05, "epoch": 1, "t": 4.241256475448608}
|
| 124 |
+
{"step": 124, "loss": 1.1066426038742065, "grad_norm": 0.512311577796936, "lr": 7.059506241219965e-05, "epoch": 1, "t": 4.8891942501068115}
|
| 125 |
+
{"step": 125, "loss": 0.9782649874687195, "grad_norm": 0.5632579326629639, "lr": 7.011034586046176e-05, "epoch": 1, "t": 4.156222105026245}
|
| 126 |
+
{"step": 126, "loss": 1.1419874429702759, "grad_norm": 0.4836714267730713, "lr": 6.962336397710819e-05, "epoch": 1, "t": 4.914880752563477}
|
| 127 |
+
{"step": 127, "loss": 1.0279699563980103, "grad_norm": 0.4782537519931793, "lr": 6.91341716182545e-05, "epoch": 1, "t": 4.600232839584351}
|
| 128 |
+
{"step": 128, "loss": 0.9862858057022095, "grad_norm": 0.6365835666656494, "lr": 6.864282388901544e-05, "epoch": 1, "t": 3.843193531036377}
|
| 129 |
+
{"step": 129, "loss": 1.039993166923523, "grad_norm": 0.5673493146896362, "lr": 6.814937613729766e-05, "epoch": 1, "t": 4.116474390029907}
|
| 130 |
+
{"step": 130, "loss": 1.1569846868515015, "grad_norm": 0.5737810730934143, "lr": 6.765388394756504e-05, "epoch": 1, "t": 5.062510967254639}
|
| 131 |
+
{"step": 131, "loss": 1.156396746635437, "grad_norm": 0.5683199167251587, "lr": 6.715640313457733e-05, "epoch": 1, "t": 5.343378782272339}
|
| 132 |
+
{"step": 132, "loss": 1.1019277572631836, "grad_norm": 0.5262781381607056, "lr": 6.665698973710288e-05, "epoch": 1, "t": 5.501410961151123}
|
| 133 |
+
{"step": 133, "loss": 1.0652830600738525, "grad_norm": 0.5279587507247925, "lr": 6.615570001160626e-05, "epoch": 1, "t": 4.568547248840332}
|
| 134 |
+
{"step": 134, "loss": 1.0896309614181519, "grad_norm": 0.5675188899040222, "lr": 6.565259042591113e-05, "epoch": 1, "t": 4.239359140396118}
|
| 135 |
+
{"step": 135, "loss": 1.145555019378662, "grad_norm": 0.5132235884666443, "lr": 6.514771765283942e-05, "epoch": 1, "t": 5.161583662033081}
|
| 136 |
+
{"step": 136, "loss": 1.1404597759246826, "grad_norm": 0.6020283102989197, "lr": 6.464113856382752e-05, "epoch": 1, "t": 4.559965133666992}
|
| 137 |
+
{"step": 137, "loss": 1.1024014949798584, "grad_norm": 0.5384811758995056, "lr": 6.413291022251989e-05, "epoch": 1, "t": 4.726520299911499}
|
| 138 |
+
{"step": 138, "loss": 1.1771068572998047, "grad_norm": 0.5445666909217834, "lr": 6.362308987834115e-05, "epoch": 1, "t": 5.2451701164245605}
|
| 139 |
+
{"step": 139, "loss": 1.0739786624908447, "grad_norm": 0.5406129360198975, "lr": 6.311173496004723e-05, "epoch": 1, "t": 4.660525321960449}
|
| 140 |
+
{"step": 140, "loss": 1.0997555255889893, "grad_norm": 0.5309702754020691, "lr": 6.259890306925627e-05, "epoch": 1, "t": 4.154837608337402}
|
| 141 |
+
{"step": 141, "loss": 0.9732502698898315, "grad_norm": 0.5423446297645569, "lr": 6.208465197396013e-05, "epoch": 1, "t": 4.477143287658691}
|
| 142 |
+
{"step": 142, "loss": 1.0984506607055664, "grad_norm": 0.5107951164245605, "lr": 6.156903960201709e-05, "epoch": 1, "t": 4.602807283401489}
|
| 143 |
+
{"step": 143, "loss": 1.0711047649383545, "grad_norm": 0.5161554217338562, "lr": 6.105212403462651e-05, "epoch": 1, "t": 4.45146107673645}
|
| 144 |
+
{"step": 144, "loss": 1.1478400230407715, "grad_norm": 0.4985423684120178, "lr": 6.0533963499786314e-05, "epoch": 1, "t": 5.01528787612915}
|
| 145 |
+
{"step": 145, "loss": 0.9466525316238403, "grad_norm": 0.5108337998390198, "lr": 6.001461636573397e-05, "epoch": 1, "t": 4.250473976135254}
|
| 146 |
+
{"step": 146, "loss": 1.0714473724365234, "grad_norm": 0.5243314504623413, "lr": 5.949414113437142e-05, "epoch": 1, "t": 4.921194791793823}
|
| 147 |
+
{"step": 147, "loss": 1.0675859451293945, "grad_norm": 0.5367836356163025, "lr": 5.897259643467527e-05, "epoch": 1, "t": 4.2278571128845215}
|
| 148 |
+
{"step": 148, "loss": 1.1217992305755615, "grad_norm": 0.6045838594436646, "lr": 5.8450041016092464e-05, "epoch": 1, "t": 4.103216171264648}
|
| 149 |
+
{"step": 149, "loss": 1.0891027450561523, "grad_norm": 0.5601292848587036, "lr": 5.792653374192245e-05, "epoch": 1, "t": 4.155539274215698}
|
| 150 |
+
{"step": 150, "loss": 1.0668423175811768, "grad_norm": 0.5978821516036987, "lr": 5.7402133582686576e-05, "epoch": 1, "t": 3.958406448364258}
|
| 151 |
+
{"step": 151, "loss": 1.1171596050262451, "grad_norm": 0.5344642400741577, "lr": 5.6876899609485256e-05, "epoch": 1, "t": 4.410540580749512}
|
| 152 |
+
{"step": 152, "loss": 1.1254163980484009, "grad_norm": 0.5476782321929932, "lr": 5.6350890987343944e-05, "epoch": 1, "t": 4.649364233016968}
|
| 153 |
+
{"step": 153, "loss": 1.088308334350586, "grad_norm": 0.48466265201568604, "lr": 5.582416696854853e-05, "epoch": 1, "t": 4.919539928436279}
|
| 154 |
+
{"step": 154, "loss": 0.9236600399017334, "grad_norm": 0.5349932909011841, "lr": 5.5296786885970805e-05, "epoch": 1, "t": 4.408169746398926}
|
| 155 |
+
{"step": 155, "loss": 1.0489084720611572, "grad_norm": 0.5243504047393799, "lr": 5.476881014638491e-05, "epoch": 1, "t": 4.4164159297943115}
|
| 156 |
+
{"step": 156, "loss": 1.17582368850708, "grad_norm": 0.5113088488578796, "lr": 5.4240296223775465e-05, "epoch": 1, "t": 4.7344701290130615}
|
| 157 |
+
{"step": 157, "loss": 1.1393662691116333, "grad_norm": 0.541817307472229, "lr": 5.3711304652638126e-05, "epoch": 1, "t": 5.64678692817688}
|
| 158 |
+
{"step": 158, "loss": 1.0807132720947266, "grad_norm": 0.6632468700408936, "lr": 5.318189502127332e-05, "epoch": 1, "t": 5.4434027671813965}
|
| 159 |
+
{"step": 159, "loss": 1.1205252408981323, "grad_norm": 0.4975724518299103, "lr": 5.265212696507387e-05, "epoch": 1, "t": 5.10357141494751}
|
| 160 |
+
{"step": 160, "loss": 1.1404213905334473, "grad_norm": 0.4893829822540283, "lr": 5.212206015980742e-05, "epoch": 1, "t": 5.291443347930908}
|
| 161 |
+
{"step": 161, "loss": 1.1716794967651367, "grad_norm": 0.579994261264801, "lr": 5.159175431489424e-05, "epoch": 1, "t": 4.521823883056641}
|
| 162 |
+
{"step": 162, "loss": 1.1746046543121338, "grad_norm": 0.5510536432266235, "lr": 5.1061269166681183e-05, "epoch": 1, "t": 4.708100318908691}
|
| 163 |
+
{"step": 163, "loss": 0.9764399528503418, "grad_norm": 0.5967614054679871, "lr": 5.053066447171282e-05, "epoch": 1, "t": 4.320418357849121}
|
| 164 |
+
{"step": 164, "loss": 1.0761276483535767, "grad_norm": 0.5031515955924988, "lr": 5e-05, "epoch": 1, "t": 4.492007255554199}
|
| 165 |
+
{"step": 165, "loss": 1.0567834377288818, "grad_norm": 0.492739200592041, "lr": 4.94693355282872e-05, "epoch": 1, "t": 5.124138116836548}
|
| 166 |
+
{"step": 166, "loss": 1.1086492538452148, "grad_norm": 0.5211777687072754, "lr": 4.893873083331882e-05, "epoch": 1, "t": 4.488057613372803}
|
| 167 |
+
{"step": 167, "loss": 1.0665874481201172, "grad_norm": 0.489278107881546, "lr": 4.840824568510579e-05, "epoch": 1, "t": 4.717162132263184}
|
| 168 |
+
{"step": 168, "loss": 0.9501603841781616, "grad_norm": 0.5504268407821655, "lr": 4.78779398401926e-05, "epoch": 1, "t": 4.038044214248657}
|
| 169 |
+
{"step": 169, "loss": 1.116373062133789, "grad_norm": 0.5695221424102783, "lr": 4.734787303492615e-05, "epoch": 1, "t": 4.288054943084717}
|
| 170 |
+
{"step": 170, "loss": 1.0904674530029297, "grad_norm": 0.5165920853614807, "lr": 4.6818104978726685e-05, "epoch": 1, "t": 4.836774587631226}
|
| 171 |
+
{"step": 171, "loss": 1.123197317123413, "grad_norm": 0.5360180735588074, "lr": 4.628869534736187e-05, "epoch": 1, "t": 4.8332624435424805}
|
| 172 |
+
{"step": 172, "loss": 0.8920274972915649, "grad_norm": 0.6180672645568848, "lr": 4.575970377622456e-05, "epoch": 1, "t": 4.7172911167144775}
|
| 173 |
+
{"step": 173, "loss": 1.0240881443023682, "grad_norm": 0.5346320271492004, "lr": 4.52311898536151e-05, "epoch": 1, "t": 4.722185373306274}
|
| 174 |
+
{"step": 174, "loss": 1.0371263027191162, "grad_norm": 0.5344886183738708, "lr": 4.47032131140292e-05, "epoch": 1, "t": 4.964730978012085}
|
| 175 |
+
{"step": 175, "loss": 1.0077505111694336, "grad_norm": 0.5254532694816589, "lr": 4.4175833031451473e-05, "epoch": 1, "t": 4.5462095737457275}
|
| 176 |
+
{"step": 176, "loss": 1.0910813808441162, "grad_norm": 0.5489541888237, "lr": 4.364910901265606e-05, "epoch": 1, "t": 4.034765958786011}
|
| 177 |
+
{"step": 177, "loss": 1.0681406259536743, "grad_norm": 0.4989105761051178, "lr": 4.3123100390514756e-05, "epoch": 1, "t": 4.542690753936768}
|
| 178 |
+
{"step": 178, "loss": 1.1236356496810913, "grad_norm": 0.55314701795578, "lr": 4.2597866417313436e-05, "epoch": 1, "t": 4.696873903274536}
|
| 179 |
+
{"step": 179, "loss": 1.101893663406372, "grad_norm": 0.5787836909294128, "lr": 4.207346625807756e-05, "epoch": 1, "t": 3.9165098667144775}
|
| 180 |
+
{"step": 180, "loss": 1.1388134956359863, "grad_norm": 0.5849209427833557, "lr": 4.1549958983907555e-05, "epoch": 1, "t": 4.51529335975647}
|
| 181 |
+
{"step": 181, "loss": 1.0979059934616089, "grad_norm": 0.5072536468505859, "lr": 4.102740356532473e-05, "epoch": 1, "t": 4.580022096633911}
|
| 182 |
+
{"step": 182, "loss": 1.0597028732299805, "grad_norm": 0.5281139612197876, "lr": 4.050585886562858e-05, "epoch": 1, "t": 4.527806043624878}
|
| 183 |
+
{"step": 183, "loss": 0.9946712255477905, "grad_norm": 0.6024081707000732, "lr": 3.998538363426605e-05, "epoch": 1, "t": 4.0228431224823}
|
| 184 |
+
{"step": 184, "loss": 1.133749008178711, "grad_norm": 0.5810251235961914, "lr": 3.94660365002137e-05, "epoch": 1, "t": 4.233653545379639}
|
| 185 |
+
{"step": 185, "loss": 1.042235255241394, "grad_norm": 0.5351488590240479, "lr": 3.894787596537352e-05, "epoch": 1, "t": 4.85040020942688}
|
| 186 |
+
{"step": 186, "loss": 1.089773416519165, "grad_norm": 0.593411386013031, "lr": 3.843096039798293e-05, "epoch": 1, "t": 4.0477845668792725}
|
| 187 |
+
{"step": 187, "loss": 1.117031216621399, "grad_norm": 0.5417437553405762, "lr": 3.791534802603988e-05, "epoch": 1, "t": 5.170712232589722}
|
| 188 |
+
{"step": 188, "loss": 1.1468424797058105, "grad_norm": 0.5026240348815918, "lr": 3.740109693074375e-05, "epoch": 1, "t": 5.059818506240845}
|
| 189 |
+
{"step": 189, "loss": 1.1194242238998413, "grad_norm": 0.6326451301574707, "lr": 3.68882650399528e-05, "epoch": 1, "t": 5.283757448196411}
|
| 190 |
+
{"step": 190, "loss": 1.1131374835968018, "grad_norm": 0.5443834066390991, "lr": 3.637691012165886e-05, "epoch": 1, "t": 5.458299160003662}
|
| 191 |
+
{"step": 191, "loss": 1.0442452430725098, "grad_norm": 0.5689224600791931, "lr": 3.586708977748012e-05, "epoch": 1, "t": 4.23190712928772}
|
| 192 |
+
{"step": 192, "loss": 1.1126645803451538, "grad_norm": 0.5516955852508545, "lr": 3.5358861436172485e-05, "epoch": 1, "t": 4.638779640197754}
|
| 193 |
+
{"step": 193, "loss": 0.9847837686538696, "grad_norm": 0.5645774602890015, "lr": 3.485228234716058e-05, "epoch": 1, "t": 4.052290916442871}
|
| 194 |
+
{"step": 194, "loss": 1.063437819480896, "grad_norm": 0.6118154525756836, "lr": 3.434740957408889e-05, "epoch": 1, "t": 4.010751724243164}
|
| 195 |
+
{"step": 195, "loss": 1.158060073852539, "grad_norm": 0.49794816970825195, "lr": 3.3844299988393755e-05, "epoch": 1, "t": 5.0552308559417725}
|
| 196 |
+
{"step": 196, "loss": 0.9808290004730225, "grad_norm": 0.6830561757087708, "lr": 3.334301026289712e-05, "epoch": 1, "t": 4.46864914894104}
|
| 197 |
+
{"step": 197, "loss": 1.0011578798294067, "grad_norm": 0.5124518871307373, "lr": 3.284359686542269e-05, "epoch": 1, "t": 4.262362241744995}
|
| 198 |
+
{"step": 198, "loss": 1.068244218826294, "grad_norm": 0.479539155960083, "lr": 3.234611605243496e-05, "epoch": 1, "t": 5.344050884246826}
|
| 199 |
+
{"step": 199, "loss": 1.0668308734893799, "grad_norm": 0.5142190456390381, "lr": 3.1850623862702344e-05, "epoch": 1, "t": 4.578876256942749}
|
| 200 |
+
{"step": 200, "loss": 1.059356927871704, "grad_norm": 0.5470719933509827, "lr": 3.135717611098458e-05, "epoch": 1, "t": 4.529844760894775}
|
| 201 |
+
{"step": 201, "loss": 1.1023340225219727, "grad_norm": 0.5097795128822327, "lr": 3.086582838174551e-05, "epoch": 1, "t": 4.714038133621216}
|
| 202 |
+
{"step": 202, "loss": 1.0163545608520508, "grad_norm": 0.507060706615448, "lr": 3.0376636022891812e-05, "epoch": 1, "t": 4.313427686691284}
|
| 203 |
+
{"step": 203, "loss": 1.1238795518875122, "grad_norm": 0.5084006786346436, "lr": 2.9889654139538246e-05, "epoch": 1, "t": 4.563612461090088}
|
| 204 |
+
{"step": 204, "loss": 1.0976316928863525, "grad_norm": 0.5512629151344299, "lr": 2.9404937587800375e-05, "epoch": 1, "t": 4.3181281089782715}
|
| 205 |
+
{"step": 205, "loss": 1.0222280025482178, "grad_norm": 0.5674687623977661, "lr": 2.8922540968615286e-05, "epoch": 1, "t": 4.1513307094573975}
|
| 206 |
+
{"step": 206, "loss": 1.027825951576233, "grad_norm": 0.5027503371238708, "lr": 2.8442518621591086e-05, "epoch": 1, "t": 4.666209936141968}
|
| 207 |
+
{"step": 207, "loss": 1.1769216060638428, "grad_norm": 0.50437331199646, "lr": 2.7964924618885778e-05, "epoch": 1, "t": 5.702998399734497}
|
| 208 |
+
{"step": 208, "loss": 1.0159366130828857, "grad_norm": 0.552759051322937, "lr": 2.748981275911633e-05, "epoch": 1, "t": 4.757322311401367}
|
| 209 |
+
{"step": 209, "loss": 1.0479300022125244, "grad_norm": 0.5230646729469299, "lr": 2.701723656129851e-05, "epoch": 1, "t": 4.330077648162842}
|
| 210 |
+
{"step": 210, "loss": 1.065136194229126, "grad_norm": 0.5290108323097229, "lr": 2.6547249258818164e-05, "epoch": 1, "t": 4.422900199890137}
|
| 211 |
+
{"step": 211, "loss": 1.0446008443832397, "grad_norm": 0.5066290497779846, "lr": 2.607990379343489e-05, "epoch": 1, "t": 4.372257232666016}
|
| 212 |
+
{"step": 212, "loss": 1.071063756942749, "grad_norm": 0.6194134950637817, "lr": 2.5615252809318284e-05, "epoch": 1, "t": 4.216610431671143}
|
| 213 |
+
{"step": 213, "loss": 1.0344080924987793, "grad_norm": 0.53545743227005, "lr": 2.5153348647117857e-05, "epoch": 1, "t": 4.531871795654297}
|
| 214 |
+
{"step": 214, "loss": 1.1218066215515137, "grad_norm": 0.536637008190155, "lr": 2.469424333806718e-05, "epoch": 1, "t": 4.433279752731323}
|
| 215 |
+
{"step": 215, "loss": 0.9499108791351318, "grad_norm": 0.5313050746917725, "lr": 2.4237988598122752e-05, "epoch": 1, "t": 4.600785493850708}
|
| 216 |
+
{"step": 216, "loss": 0.9986656308174133, "grad_norm": 0.5128288269042969, "lr": 2.3784635822138424e-05, "epoch": 1, "t": 4.526764154434204}
|
| 217 |
+
{"step": 217, "loss": 0.9684886336326599, "grad_norm": 0.5480437874794006, "lr": 2.333423607807613e-05, "epoch": 1, "t": 3.9635729789733887}
|
| 218 |
+
{"step": 218, "loss": 1.069244384765625, "grad_norm": 0.5504754781723022, "lr": 2.288684010125325e-05, "epoch": 1, "t": 4.310101747512817}
|
| 219 |
+
{"step": 219, "loss": 1.022673487663269, "grad_norm": 0.6144146919250488, "lr": 2.2442498288627556e-05, "epoch": 1, "t": 4.083557367324829}
|
| 220 |
+
{"step": 220, "loss": 1.079294204711914, "grad_norm": 0.5259053707122803, "lr": 2.2001260693120233e-05, "epoch": 1, "t": 4.832204818725586}
|
| 221 |
+
{"step": 221, "loss": 1.1039211750030518, "grad_norm": 0.5115348696708679, "lr": 2.156317701797766e-05, "epoch": 1, "t": 4.994130373001099}
|
| 222 |
+
{"step": 222, "loss": 1.0928882360458374, "grad_norm": 0.5184217691421509, "lr": 2.1128296611172593e-05, "epoch": 1, "t": 5.0237274169921875}
|
| 223 |
+
{"step": 223, "loss": 1.005937099456787, "grad_norm": 0.4899376928806305, "lr": 2.0696668459845355e-05, "epoch": 1, "t": 5.284451007843018}
|
| 224 |
+
{"step": 224, "loss": 0.9694923162460327, "grad_norm": 0.5294512510299683, "lr": 2.026834118478567e-05, "epoch": 1, "t": 4.282395601272583}
|
| 225 |
+
{"step": 225, "loss": 1.0955185890197754, "grad_norm": 0.5702258348464966, "lr": 1.98433630349558e-05, "epoch": 1, "t": 4.199642658233643}
|
| 226 |
+
{"step": 226, "loss": 1.0124088525772095, "grad_norm": 0.4795595407485962, "lr": 1.9421781882055444e-05, "epoch": 1, "t": 5.957972049713135}
|
| 227 |
+
{"step": 227, "loss": 1.1133861541748047, "grad_norm": 0.4973354637622833, "lr": 1.9003645215129355e-05, "epoch": 1, "t": 5.491847276687622}
|
| 228 |
+
{"step": 228, "loss": 1.0406187772750854, "grad_norm": 0.5038388967514038, "lr": 1.858900013521788e-05, "epoch": 1, "t": 4.84385871887207}
|
| 229 |
+
{"step": 229, "loss": 1.1676573753356934, "grad_norm": 0.544230580329895, "lr": 1.817789335005121e-05, "epoch": 1, "t": 4.261170387268066}
|
| 230 |
+
{"step": 230, "loss": 1.0059478282928467, "grad_norm": 0.5692961812019348, "lr": 1.777037116878804e-05, "epoch": 1, "t": 4.290574550628662}
|
| 231 |
+
{"step": 231, "loss": 0.9805676341056824, "grad_norm": 0.6133225560188293, "lr": 1.7366479496799077e-05, "epoch": 1, "t": 4.273016929626465}
|
| 232 |
+
{"step": 232, "loss": 1.0549302101135254, "grad_norm": 0.5254215002059937, "lr": 1.6966263830495936e-05, "epoch": 1, "t": 4.395004987716675}
|
| 233 |
+
{"step": 233, "loss": 1.082611083984375, "grad_norm": 0.5463281869888306, "lr": 1.656976925220633e-05, "epoch": 1, "t": 4.464715480804443}
|
| 234 |
+
{"step": 234, "loss": 1.1121950149536133, "grad_norm": 0.5425562858581543, "lr": 1.6177040425095662e-05, "epoch": 1, "t": 4.525009393692017}
|
| 235 |
+
{"step": 235, "loss": 1.1501713991165161, "grad_norm": 0.5191682577133179, "lr": 1.5788121588135975e-05, "epoch": 1, "t": 4.914149284362793}
|
| 236 |
+
{"step": 236, "loss": 1.1118168830871582, "grad_norm": 0.545390784740448, "lr": 1.5403056551122697e-05, "epoch": 1, "t": 4.637571096420288}
|
| 237 |
+
{"step": 237, "loss": 1.0449159145355225, "grad_norm": 0.595125675201416, "lr": 1.5021888689739549e-05, "epoch": 1, "t": 4.35769248008728}
|
| 238 |
+
{"step": 238, "loss": 0.867507815361023, "grad_norm": 0.5165945291519165, "lr": 1.4644660940672627e-05, "epoch": 1, "t": 4.313535213470459}
|
| 239 |
+
{"step": 239, "loss": 1.0244340896606445, "grad_norm": 0.6335296630859375, "lr": 1.427141579677374e-05, "epoch": 1, "t": 4.4786293506622314}
|
| 240 |
+
{"step": 240, "loss": 0.9630542397499084, "grad_norm": 0.5498972535133362, "lr": 1.3902195302273779e-05, "epoch": 1, "t": 4.027124881744385}
|
| 241 |
+
{"step": 241, "loss": 1.051306962966919, "grad_norm": 0.5721133351325989, "lr": 1.3537041048046695e-05, "epoch": 1, "t": 4.3420069217681885}
|
| 242 |
+
{"step": 242, "loss": 0.8949457406997681, "grad_norm": 0.5284407734870911, "lr": 1.3175994166924394e-05, "epoch": 1, "t": 4.92367696762085}
|
| 243 |
+
{"step": 243, "loss": 1.0572562217712402, "grad_norm": 0.6526340842247009, "lr": 1.2819095329063469e-05, "epoch": 1, "t": 4.9920361042022705}
|
| 244 |
+
{"step": 244, "loss": 1.041996717453003, "grad_norm": 0.5489502549171448, "lr": 1.246638473736378e-05, "epoch": 1, "t": 4.186185359954834}
|
| 245 |
+
{"step": 245, "loss": 1.022842526435852, "grad_norm": 0.6071504950523376, "lr": 1.2117902122939861e-05, "epoch": 1, "t": 3.8784339427948}
|
| 246 |
+
{"step": 246, "loss": 1.112472414970398, "grad_norm": 0.5546125173568726, "lr": 1.1773686740645384e-05, "epoch": 1, "t": 4.28344464302063}
|
| 247 |
+
{"step": 247, "loss": 1.0369735956192017, "grad_norm": 0.5636987090110779, "lr": 1.1433777364651271e-05, "epoch": 1, "t": 4.092544317245483}
|
| 248 |
+
{"step": 248, "loss": 0.982390284538269, "grad_norm": 0.5377970337867737, "lr": 1.1098212284078036e-05, "epoch": 1, "t": 4.842264175415039}
|
| 249 |
+
{"step": 249, "loss": 0.954643726348877, "grad_norm": 0.6366534233093262, "lr": 1.076702929868264e-05, "epoch": 1, "t": 4.775829076766968}
|
| 250 |
+
{"step": 250, "loss": 1.18538236618042, "grad_norm": 0.48834678530693054, "lr": 1.0440265714600572e-05, "epoch": 1, "t": 5.253664970397949}
|
| 251 |
+
{"step": 251, "loss": 1.106203556060791, "grad_norm": 0.6136064529418945, "lr": 1.0117958340143507e-05, "epoch": 1, "t": 4.804788827896118}
|
| 252 |
+
{"step": 252, "loss": 0.9662821292877197, "grad_norm": 0.5683513283729553, "lr": 9.800143481652979e-06, "epoch": 1, "t": 4.521551132202148}
|
| 253 |
+
{"step": 253, "loss": 1.1467392444610596, "grad_norm": 0.5336821675300598, "lr": 9.48685693941067e-06, "epoch": 1, "t": 4.732549428939819}
|
| 254 |
+
{"step": 254, "loss": 1.0444958209991455, "grad_norm": 0.5080744624137878, "lr": 9.17813400360572e-06, "epoch": 1, "t": 4.868363857269287}
|
| 255 |
+
{"step": 255, "loss": 1.122852087020874, "grad_norm": 0.49413320422172546, "lr": 8.874009450359427e-06, "epoch": 1, "t": 4.720521926879883}
|
| 256 |
+
{"step": 256, "loss": 1.0515964031219482, "grad_norm": 0.6242619156837463, "lr": 8.574517537807897e-06, "epoch": 1, "t": 4.521535634994507}
|
| 257 |
+
{"step": 257, "loss": 1.129776954650879, "grad_norm": 0.5735353827476501, "lr": 8.279692002243027e-06, "epoch": 1, "t": 4.3324010372161865}
|
| 258 |
+
{"step": 258, "loss": 1.0192005634307861, "grad_norm": 0.5799077153205872, "lr": 7.989566054312287e-06, "epoch": 1, "t": 4.0833210945129395}
|
| 259 |
+
{"step": 259, "loss": 1.074305534362793, "grad_norm": 0.5495190024375916, "lr": 7.704172375277691e-06, "epoch": 1, "t": 4.3436784744262695}
|
| 260 |
+
{"step": 260, "loss": 1.0794835090637207, "grad_norm": 0.5795458555221558, "lr": 7.423543113334436e-06, "epoch": 1, "t": 4.735230207443237}
|
| 261 |
+
{"step": 261, "loss": 1.0177257061004639, "grad_norm": 1.0287628173828125, "lr": 7.14770987998954e-06, "epoch": 1, "t": 4.3184332847595215}
|
| 262 |
+
{"step": 262, "loss": 0.828809380531311, "grad_norm": 0.5874334573745728, "lr": 6.876703746500984e-06, "epoch": 1, "t": 3.8442461490631104}
|
| 263 |
+
{"step": 263, "loss": 1.0659030675888062, "grad_norm": 0.5218153595924377, "lr": 6.610555240377652e-06, "epoch": 1, "t": 4.4946722984313965}
|
| 264 |
+
{"step": 264, "loss": 0.95011305809021, "grad_norm": 0.5335944294929504, "lr": 6.349294341940593e-06, "epoch": 1, "t": 4.286647796630859}
|
| 265 |
+
{"step": 265, "loss": 1.1452821493148804, "grad_norm": 0.5478469133377075, "lr": 6.092950480945897e-06, "epoch": 1, "t": 4.722761631011963}
|
| 266 |
+
{"step": 266, "loss": 1.0393742322921753, "grad_norm": 0.5592689514160156, "lr": 5.841552533269534e-06, "epoch": 1, "t": 4.275075912475586}
|
| 267 |
+
{"step": 267, "loss": 1.03959321975708, "grad_norm": 0.5026338696479797, "lr": 5.595128817654638e-06, "epoch": 1, "t": 4.761636972427368}
|
| 268 |
+
{"step": 268, "loss": 1.1046674251556396, "grad_norm": 0.6093679666519165, "lr": 5.353707092521582e-06, "epoch": 1, "t": 4.564788579940796}
|
| 269 |
+
{"step": 269, "loss": 1.0807722806930542, "grad_norm": 0.5893504023551941, "lr": 5.117314552841052e-06, "epoch": 1, "t": 4.425111293792725}
|
| 270 |
+
{"step": 270, "loss": 1.0272538661956787, "grad_norm": 0.6843682527542114, "lr": 4.885977827070748e-06, "epoch": 1, "t": 4.529506683349609}
|
| 271 |
+
{"step": 271, "loss": 1.0035436153411865, "grad_norm": 0.5748874545097351, "lr": 4.659722974155767e-06, "epoch": 1, "t": 3.99251389503479}
|
| 272 |
+
{"step": 272, "loss": 0.9947246313095093, "grad_norm": 0.5380876064300537, "lr": 4.43857548059321e-06, "epoch": 1, "t": 4.604004621505737}
|
| 273 |
+
{"step": 273, "loss": 1.0328400135040283, "grad_norm": 0.5722586512565613, "lr": 4.2225602575612755e-06, "epoch": 1, "t": 4.030544757843018}
|
| 274 |
+
{"step": 274, "loss": 1.0134365558624268, "grad_norm": 0.4946887195110321, "lr": 4.011701638113063e-06, "epoch": 1, "t": 4.405919075012207}
|
| 275 |
+
{"step": 275, "loss": 1.0163841247558594, "grad_norm": 0.5077712535858154, "lr": 3.8060233744356633e-06, "epoch": 1, "t": 4.599614143371582}
|
| 276 |
+
{"step": 276, "loss": 1.0606365203857422, "grad_norm": 0.5821368098258972, "lr": 3.605548635174533e-06, "epoch": 1, "t": 4.243448495864868}
|
| 277 |
+
{"step": 277, "loss": 1.109765648841858, "grad_norm": 0.5064703822135925, "lr": 3.410300002823691e-06, "epoch": 1, "t": 5.325015544891357}
|
| 278 |
+
{"step": 278, "loss": 1.0385165214538574, "grad_norm": 0.5606745481491089, "lr": 3.220299471181898e-06, "epoch": 1, "t": 4.729890584945679}
|
| 279 |
+
{"step": 279, "loss": 1.058180332183838, "grad_norm": 0.5086253881454468, "lr": 3.035568442875136e-06, "epoch": 1, "t": 4.912935733795166}
|
| 280 |
+
{"step": 280, "loss": 1.07375967502594, "grad_norm": 0.5348255634307861, "lr": 2.85612772694579e-06, "epoch": 1, "t": 4.427133083343506}
|
| 281 |
+
{"step": 281, "loss": 0.9372953176498413, "grad_norm": 0.4740752875804901, "lr": 2.6819975365085237e-06, "epoch": 1, "t": 4.645989179611206}
|
| 282 |
+
{"step": 282, "loss": 1.0597269535064697, "grad_norm": 0.6335656642913818, "lr": 2.5131974864734066e-06, "epoch": 1, "t": 4.064257621765137}
|
| 283 |
+
{"step": 283, "loss": 1.0182781219482422, "grad_norm": 0.5961326956748962, "lr": 2.349746591336405e-06, "epoch": 1, "t": 4.167037010192871}
|
| 284 |
+
{"step": 284, "loss": 1.0584691762924194, "grad_norm": 0.5399417281150818, "lr": 2.191663263037458e-06, "epoch": 1, "t": 4.3510825634002686}
|
| 285 |
+
{"step": 285, "loss": 1.1428117752075195, "grad_norm": 0.5469995737075806, "lr": 2.0389653088865036e-06, "epoch": 1, "t": 4.520026206970215}
|
| 286 |
+
{"step": 286, "loss": 1.0124881267547607, "grad_norm": 0.5254855751991272, "lr": 1.8916699295575324e-06, "epoch": 1, "t": 4.595729351043701}
|
| 287 |
+
{"step": 287, "loss": 1.0624876022338867, "grad_norm": 0.5347287654876709, "lr": 1.7497937171510547e-06, "epoch": 1, "t": 4.592611312866211}
|
| 288 |
+
{"step": 288, "loss": 1.0069644451141357, "grad_norm": 0.5161753296852112, "lr": 1.6133526533250565e-06, "epoch": 1, "t": 4.45608115196228}
|
| 289 |
+
{"step": 289, "loss": 1.1297398805618286, "grad_norm": 1.414000153541565, "lr": 1.4823621074947503e-06, "epoch": 1, "t": 4.3639490604400635}
|
| 290 |
+
{"step": 290, "loss": 0.9976916313171387, "grad_norm": 0.48237261176109314, "lr": 1.3568368351012717e-06, "epoch": 1, "t": 4.7519214153289795}
|
| 291 |
+
{"step": 291, "loss": 1.1876258850097656, "grad_norm": 0.4967099130153656, "lr": 1.236790975949592e-06, "epoch": 1, "t": 5.402672290802002}
|
| 292 |
+
{"step": 292, "loss": 1.134186029434204, "grad_norm": 0.6165451407432556, "lr": 1.1222380526156928e-06, "epoch": 1, "t": 4.552279472351074}
|
| 293 |
+
{"step": 293, "loss": 1.0102145671844482, "grad_norm": 0.5769509673118591, "lr": 1.0131909689233442e-06, "epoch": 1, "t": 5.032619476318359}
|
| 294 |
+
{"step": 294, "loss": 1.00609290599823, "grad_norm": 0.6198142170906067, "lr": 9.096620084905472e-07, "epoch": 1, "t": 4.278170108795166}
|
| 295 |
+
{"step": 295, "loss": 1.0029926300048828, "grad_norm": 0.5119993686676025, "lr": 8.11662833345822e-07, "epoch": 1, "t": 4.514641761779785}
|
| 296 |
+
{"step": 296, "loss": 1.103434443473816, "grad_norm": 0.573027491569519, "lr": 7.192044826145771e-07, "epoch": 1, "t": 4.448399066925049}
|
| 297 |
+
{"step": 297, "loss": 1.071667194366455, "grad_norm": 0.577226459980011, "lr": 6.322973712755697e-07, "epoch": 1, "t": 4.744686841964722}
|
| 298 |
+
{"step": 298, "loss": 1.0070140361785889, "grad_norm": 0.5720388293266296, "lr": 5.509512889877333e-07, "epoch": 1, "t": 4.5218751430511475}
|
| 299 |
+
{"step": 299, "loss": 1.0531494617462158, "grad_norm": 0.6200311779975891, "lr": 4.7517539898741524e-07, "epoch": 1, "t": 4.783984422683716}
|
| 300 |
+
{"step": 300, "loss": 1.031361699104309, "grad_norm": 0.4972401559352875, "lr": 4.049782370561583e-07, "epoch": 1, "t": 4.783889532089233}
|
| 301 |
+
{"step": 301, "loss": 0.9963631629943848, "grad_norm": 0.50146484375, "lr": 3.4036771055923066e-07, "epoch": 1, "t": 4.60985803604126}
|
| 302 |
+
{"step": 302, "loss": 0.986006498336792, "grad_norm": 0.5622588992118835, "lr": 2.813510975548772e-07, "epoch": 1, "t": 4.139053106307983}
|
| 303 |
+
{"step": 303, "loss": 1.0686516761779785, "grad_norm": 0.4927791953086853, "lr": 2.2793504597447002e-07, "epoch": 1, "t": 4.635111093521118}
|
| 304 |
+
{"step": 304, "loss": 1.091249942779541, "grad_norm": 0.517026960849762, "lr": 1.8012557287367392e-07, "epoch": 1, "t": 4.854142189025879}
|
| 305 |
+
{"step": 305, "loss": 1.1145005226135254, "grad_norm": 0.5055581331253052, "lr": 1.379280637546443e-07, "epoch": 1, "t": 4.547342777252197}
|
| 306 |
+
{"step": 306, "loss": 1.0295354127883911, "grad_norm": 0.49713513255119324, "lr": 1.0134727195937333e-07, "epoch": 1, "t": 5.182791471481323}
|
| 307 |
+
{"step": 307, "loss": 0.9750290513038635, "grad_norm": 0.5073899626731873, "lr": 7.038731813426291e-08, "epoch": 1, "t": 5.048128604888916}
|
| 308 |
+
{"step": 308, "loss": 1.188929796218872, "grad_norm": 0.5364013314247131, "lr": 4.5051689765929214e-08, "epoch": 1, "t": 5.233379602432251}
|
| 309 |
+
{"step": 309, "loss": 1.0469988584518433, "grad_norm": 0.5482891798019409, "lr": 2.534324078837802e-08, "epoch": 1, "t": 4.223793029785156}
|
| 310 |
+
{"step": 310, "loss": 1.120493769645691, "grad_norm": 0.5453365445137024, "lr": 1.1264191261528557e-08, "epoch": 1, "t": 4.377922534942627}
|
| 311 |
+
{"step": 311, "loss": 0.9584072828292847, "grad_norm": 0.5121012926101685, "lr": 2.8161271211024633e-09, "epoch": 1, "t": 4.6319379806518555}
|
| 312 |
+
{"step": 312, "loss": 1.0436558723449707, "grad_norm": 0.5431374311447144, "lr": 0.0, "epoch": 1, "t": 4.32390284538269}
|
gemma-3-27b-it/top-s0/README.md
ADDED
|
@@ -0,0 +1,206 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: google/gemma-3-27b-it
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- base_model:adapter:google/gemma-3-27b-it
|
| 6 |
+
- lora
|
| 7 |
+
- transformers
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# Model Card for Model ID
|
| 11 |
+
|
| 12 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
## Model Details
|
| 17 |
+
|
| 18 |
+
### Model Description
|
| 19 |
+
|
| 20 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
- **Developed by:** [More Information Needed]
|
| 25 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 26 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 27 |
+
- **Model type:** [More Information Needed]
|
| 28 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 29 |
+
- **License:** [More Information Needed]
|
| 30 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 31 |
+
|
| 32 |
+
### Model Sources [optional]
|
| 33 |
+
|
| 34 |
+
<!-- Provide the basic links for the model. -->
|
| 35 |
+
|
| 36 |
+
- **Repository:** [More Information Needed]
|
| 37 |
+
- **Paper [optional]:** [More Information Needed]
|
| 38 |
+
- **Demo [optional]:** [More Information Needed]
|
| 39 |
+
|
| 40 |
+
## Uses
|
| 41 |
+
|
| 42 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 43 |
+
|
| 44 |
+
### Direct Use
|
| 45 |
+
|
| 46 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 47 |
+
|
| 48 |
+
[More Information Needed]
|
| 49 |
+
|
| 50 |
+
### Downstream Use [optional]
|
| 51 |
+
|
| 52 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 53 |
+
|
| 54 |
+
[More Information Needed]
|
| 55 |
+
|
| 56 |
+
### Out-of-Scope Use
|
| 57 |
+
|
| 58 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 59 |
+
|
| 60 |
+
[More Information Needed]
|
| 61 |
+
|
| 62 |
+
## Bias, Risks, and Limitations
|
| 63 |
+
|
| 64 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 65 |
+
|
| 66 |
+
[More Information Needed]
|
| 67 |
+
|
| 68 |
+
### Recommendations
|
| 69 |
+
|
| 70 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 71 |
+
|
| 72 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 73 |
+
|
| 74 |
+
## How to Get Started with the Model
|
| 75 |
+
|
| 76 |
+
Use the code below to get started with the model.
|
| 77 |
+
|
| 78 |
+
[More Information Needed]
|
| 79 |
+
|
| 80 |
+
## Training Details
|
| 81 |
+
|
| 82 |
+
### Training Data
|
| 83 |
+
|
| 84 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 85 |
+
|
| 86 |
+
[More Information Needed]
|
| 87 |
+
|
| 88 |
+
### Training Procedure
|
| 89 |
+
|
| 90 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 91 |
+
|
| 92 |
+
#### Preprocessing [optional]
|
| 93 |
+
|
| 94 |
+
[More Information Needed]
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
#### Training Hyperparameters
|
| 98 |
+
|
| 99 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 100 |
+
|
| 101 |
+
#### Speeds, Sizes, Times [optional]
|
| 102 |
+
|
| 103 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 104 |
+
|
| 105 |
+
[More Information Needed]
|
| 106 |
+
|
| 107 |
+
## Evaluation
|
| 108 |
+
|
| 109 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 110 |
+
|
| 111 |
+
### Testing Data, Factors & Metrics
|
| 112 |
+
|
| 113 |
+
#### Testing Data
|
| 114 |
+
|
| 115 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 116 |
+
|
| 117 |
+
[More Information Needed]
|
| 118 |
+
|
| 119 |
+
#### Factors
|
| 120 |
+
|
| 121 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 122 |
+
|
| 123 |
+
[More Information Needed]
|
| 124 |
+
|
| 125 |
+
#### Metrics
|
| 126 |
+
|
| 127 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 128 |
+
|
| 129 |
+
[More Information Needed]
|
| 130 |
+
|
| 131 |
+
### Results
|
| 132 |
+
|
| 133 |
+
[More Information Needed]
|
| 134 |
+
|
| 135 |
+
#### Summary
|
| 136 |
+
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
## Model Examination [optional]
|
| 140 |
+
|
| 141 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 142 |
+
|
| 143 |
+
[More Information Needed]
|
| 144 |
+
|
| 145 |
+
## Environmental Impact
|
| 146 |
+
|
| 147 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 148 |
+
|
| 149 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 150 |
+
|
| 151 |
+
- **Hardware Type:** [More Information Needed]
|
| 152 |
+
- **Hours used:** [More Information Needed]
|
| 153 |
+
- **Cloud Provider:** [More Information Needed]
|
| 154 |
+
- **Compute Region:** [More Information Needed]
|
| 155 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 156 |
+
|
| 157 |
+
## Technical Specifications [optional]
|
| 158 |
+
|
| 159 |
+
### Model Architecture and Objective
|
| 160 |
+
|
| 161 |
+
[More Information Needed]
|
| 162 |
+
|
| 163 |
+
### Compute Infrastructure
|
| 164 |
+
|
| 165 |
+
[More Information Needed]
|
| 166 |
+
|
| 167 |
+
#### Hardware
|
| 168 |
+
|
| 169 |
+
[More Information Needed]
|
| 170 |
+
|
| 171 |
+
#### Software
|
| 172 |
+
|
| 173 |
+
[More Information Needed]
|
| 174 |
+
|
| 175 |
+
## Citation [optional]
|
| 176 |
+
|
| 177 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 178 |
+
|
| 179 |
+
**BibTeX:**
|
| 180 |
+
|
| 181 |
+
[More Information Needed]
|
| 182 |
+
|
| 183 |
+
**APA:**
|
| 184 |
+
|
| 185 |
+
[More Information Needed]
|
| 186 |
+
|
| 187 |
+
## Glossary [optional]
|
| 188 |
+
|
| 189 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 190 |
+
|
| 191 |
+
[More Information Needed]
|
| 192 |
+
|
| 193 |
+
## More Information [optional]
|
| 194 |
+
|
| 195 |
+
[More Information Needed]
|
| 196 |
+
|
| 197 |
+
## Model Card Authors [optional]
|
| 198 |
+
|
| 199 |
+
[More Information Needed]
|
| 200 |
+
|
| 201 |
+
## Model Card Contact
|
| 202 |
+
|
| 203 |
+
[More Information Needed]
|
| 204 |
+
### Framework versions
|
| 205 |
+
|
| 206 |
+
- PEFT 0.17.1
|
gemma-3-27b-it/top-s0/adapter_config.json
ADDED
|
@@ -0,0 +1,228 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": {
|
| 4 |
+
"base_model_class": "Gemma3ForConditionalGeneration",
|
| 5 |
+
"parent_library": "transformers.models.gemma3.modeling_gemma3"
|
| 6 |
+
},
|
| 7 |
+
"base_model_name_or_path": "google/gemma-3-27b-it",
|
| 8 |
+
"bias": "none",
|
| 9 |
+
"corda_config": null,
|
| 10 |
+
"eva_config": null,
|
| 11 |
+
"exclude_modules": null,
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": true,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layer_replication": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"layers_to_transform": null,
|
| 18 |
+
"loftq_config": {},
|
| 19 |
+
"lora_alpha": 128,
|
| 20 |
+
"lora_bias": false,
|
| 21 |
+
"lora_dropout": 0.05,
|
| 22 |
+
"megatron_config": null,
|
| 23 |
+
"megatron_core": "megatron.core",
|
| 24 |
+
"modules_to_save": null,
|
| 25 |
+
"peft_type": "LORA",
|
| 26 |
+
"qalora_group_size": 16,
|
| 27 |
+
"r": 64,
|
| 28 |
+
"rank_pattern": {},
|
| 29 |
+
"revision": null,
|
| 30 |
+
"target_modules": [
|
| 31 |
+
"40.self_attn.k_proj",
|
| 32 |
+
"28.self_attn.v_proj",
|
| 33 |
+
"35.self_attn.q_proj",
|
| 34 |
+
"language_model.layers.4.self_attn.v_proj",
|
| 35 |
+
"58.self_attn.q_proj",
|
| 36 |
+
"language_model.layers.14.self_attn.q_proj",
|
| 37 |
+
"language_model.layers.3.self_attn.v_proj",
|
| 38 |
+
"language_model.layers.20.self_attn.k_proj",
|
| 39 |
+
"38.self_attn.k_proj",
|
| 40 |
+
"33.self_attn.k_proj",
|
| 41 |
+
"language_model.layers.24.self_attn.q_proj",
|
| 42 |
+
"language_model.layers.18.self_attn.q_proj",
|
| 43 |
+
"34.self_attn.k_proj",
|
| 44 |
+
"33.self_attn.q_proj",
|
| 45 |
+
"language_model.layers.23.self_attn.k_proj",
|
| 46 |
+
"language_model.layers.8.self_attn.v_proj",
|
| 47 |
+
"language_model.layers.7.self_attn.q_proj",
|
| 48 |
+
"32.self_attn.v_proj",
|
| 49 |
+
"language_model.layers.11.self_attn.q_proj",
|
| 50 |
+
"language_model.layers.21.self_attn.q_proj",
|
| 51 |
+
"language_model.layers.14.self_attn.k_proj",
|
| 52 |
+
"58.self_attn.v_proj",
|
| 53 |
+
"53.self_attn.q_proj",
|
| 54 |
+
"47.self_attn.v_proj",
|
| 55 |
+
"45.self_attn.v_proj",
|
| 56 |
+
"34.self_attn.q_proj",
|
| 57 |
+
"language_model.layers.3.self_attn.q_proj",
|
| 58 |
+
"36.self_attn.k_proj",
|
| 59 |
+
"51.self_attn.k_proj",
|
| 60 |
+
"54.self_attn.v_proj",
|
| 61 |
+
"language_model.layers.13.self_attn.q_proj",
|
| 62 |
+
"language_model.layers.10.self_attn.q_proj",
|
| 63 |
+
"58.self_attn.k_proj",
|
| 64 |
+
"language_model.layers.12.self_attn.q_proj",
|
| 65 |
+
"46.self_attn.k_proj",
|
| 66 |
+
"41.self_attn.v_proj",
|
| 67 |
+
"52.self_attn.k_proj",
|
| 68 |
+
"language_model.layers.6.self_attn.k_proj",
|
| 69 |
+
"language_model.layers.12.self_attn.k_proj",
|
| 70 |
+
"language_model.layers.0.self_attn.q_proj",
|
| 71 |
+
"44.self_attn.q_proj",
|
| 72 |
+
"language_model.layers.9.self_attn.k_proj",
|
| 73 |
+
"language_model.layers.7.self_attn.v_proj",
|
| 74 |
+
"61.self_attn.k_proj",
|
| 75 |
+
"28.self_attn.q_proj",
|
| 76 |
+
"language_model.layers.19.self_attn.k_proj",
|
| 77 |
+
"language_model.layers.22.self_attn.v_proj",
|
| 78 |
+
"language_model.layers.1.self_attn.k_proj",
|
| 79 |
+
"language_model.layers.17.self_attn.q_proj",
|
| 80 |
+
"language_model.layers.5.self_attn.v_proj",
|
| 81 |
+
"30.self_attn.v_proj",
|
| 82 |
+
"language_model.layers.6.self_attn.v_proj",
|
| 83 |
+
"language_model.layers.11.self_attn.k_proj",
|
| 84 |
+
"41.self_attn.k_proj",
|
| 85 |
+
"language_model.layers.10.self_attn.k_proj",
|
| 86 |
+
"38.self_attn.v_proj",
|
| 87 |
+
"language_model.layers.26.self_attn.k_proj",
|
| 88 |
+
"language_model.layers.25.self_attn.k_proj",
|
| 89 |
+
"language_model.layers.8.self_attn.k_proj",
|
| 90 |
+
"language_model.layers.23.self_attn.v_proj",
|
| 91 |
+
"50.self_attn.v_proj",
|
| 92 |
+
"language_model.layers.20.self_attn.q_proj",
|
| 93 |
+
"57.self_attn.k_proj",
|
| 94 |
+
"language_model.layers.15.self_attn.k_proj",
|
| 95 |
+
"language_model.layers.21.self_attn.k_proj",
|
| 96 |
+
"37.self_attn.k_proj",
|
| 97 |
+
"language_model.layers.16.self_attn.q_proj",
|
| 98 |
+
"language_model.layers.0.self_attn.v_proj",
|
| 99 |
+
"53.self_attn.k_proj",
|
| 100 |
+
"language_model.layers.13.self_attn.k_proj",
|
| 101 |
+
"language_model.layers.24.self_attn.v_proj",
|
| 102 |
+
"27.self_attn.v_proj",
|
| 103 |
+
"language_model.layers.2.self_attn.k_proj",
|
| 104 |
+
"47.self_attn.q_proj",
|
| 105 |
+
"60.self_attn.k_proj",
|
| 106 |
+
"language_model.layers.22.self_attn.q_proj",
|
| 107 |
+
"language_model.layers.9.self_attn.v_proj",
|
| 108 |
+
"47.self_attn.k_proj",
|
| 109 |
+
"50.self_attn.k_proj",
|
| 110 |
+
"language_model.layers.19.self_attn.v_proj",
|
| 111 |
+
"language_model.layers.15.self_attn.q_proj",
|
| 112 |
+
"language_model.layers.17.self_attn.k_proj",
|
| 113 |
+
"48.self_attn.q_proj",
|
| 114 |
+
"36.self_attn.q_proj",
|
| 115 |
+
"language_model.layers.7.self_attn.k_proj",
|
| 116 |
+
"45.self_attn.k_proj",
|
| 117 |
+
"gate_proj",
|
| 118 |
+
"language_model.layers.15.self_attn.v_proj",
|
| 119 |
+
"language_model.layers.5.self_attn.q_proj",
|
| 120 |
+
"45.self_attn.q_proj",
|
| 121 |
+
"32.self_attn.q_proj",
|
| 122 |
+
"31.self_attn.k_proj",
|
| 123 |
+
"language_model.layers.19.self_attn.q_proj",
|
| 124 |
+
"51.self_attn.v_proj",
|
| 125 |
+
"language_model.layers.2.self_attn.v_proj",
|
| 126 |
+
"30.self_attn.q_proj",
|
| 127 |
+
"language_model.layers.11.self_attn.v_proj",
|
| 128 |
+
"language_model.layers.5.self_attn.k_proj",
|
| 129 |
+
"language_model.layers.16.self_attn.k_proj",
|
| 130 |
+
"39.self_attn.v_proj",
|
| 131 |
+
"56.self_attn.v_proj",
|
| 132 |
+
"39.self_attn.k_proj",
|
| 133 |
+
"50.self_attn.q_proj",
|
| 134 |
+
"language_model.layers.2.self_attn.q_proj",
|
| 135 |
+
"language_model.layers.23.self_attn.q_proj",
|
| 136 |
+
"44.self_attn.v_proj",
|
| 137 |
+
"27.self_attn.k_proj",
|
| 138 |
+
"39.self_attn.q_proj",
|
| 139 |
+
"48.self_attn.k_proj",
|
| 140 |
+
"59.self_attn.q_proj",
|
| 141 |
+
"language_model.layers.4.self_attn.k_proj",
|
| 142 |
+
"37.self_attn.q_proj",
|
| 143 |
+
"language_model.layers.25.self_attn.q_proj",
|
| 144 |
+
"31.self_attn.v_proj",
|
| 145 |
+
"32.self_attn.k_proj",
|
| 146 |
+
"language_model.layers.4.self_attn.q_proj",
|
| 147 |
+
"44.self_attn.k_proj",
|
| 148 |
+
"53.self_attn.v_proj",
|
| 149 |
+
"language_model.layers.22.self_attn.k_proj",
|
| 150 |
+
"56.self_attn.k_proj",
|
| 151 |
+
"27.self_attn.q_proj",
|
| 152 |
+
"40.self_attn.v_proj",
|
| 153 |
+
"28.self_attn.k_proj",
|
| 154 |
+
"language_model.layers.16.self_attn.v_proj",
|
| 155 |
+
"42.self_attn.v_proj",
|
| 156 |
+
"29.self_attn.q_proj",
|
| 157 |
+
"57.self_attn.q_proj",
|
| 158 |
+
"language_model.layers.3.self_attn.k_proj",
|
| 159 |
+
"36.self_attn.v_proj",
|
| 160 |
+
"language_model.layers.20.self_attn.v_proj",
|
| 161 |
+
"language_model.layers.17.self_attn.v_proj",
|
| 162 |
+
"language_model.layers.12.self_attn.v_proj",
|
| 163 |
+
"34.self_attn.v_proj",
|
| 164 |
+
"61.self_attn.q_proj",
|
| 165 |
+
"60.self_attn.q_proj",
|
| 166 |
+
"40.self_attn.q_proj",
|
| 167 |
+
"language_model.layers.10.self_attn.v_proj",
|
| 168 |
+
"60.self_attn.v_proj",
|
| 169 |
+
"o_proj",
|
| 170 |
+
"33.self_attn.v_proj",
|
| 171 |
+
"29.self_attn.v_proj",
|
| 172 |
+
"language_model.layers.9.self_attn.q_proj",
|
| 173 |
+
"language_model.layers.26.self_attn.v_proj",
|
| 174 |
+
"language_model.layers.24.self_attn.k_proj",
|
| 175 |
+
"59.self_attn.k_proj",
|
| 176 |
+
"42.self_attn.k_proj",
|
| 177 |
+
"language_model.layers.18.self_attn.v_proj",
|
| 178 |
+
"55.self_attn.v_proj",
|
| 179 |
+
"language_model.layers.26.self_attn.q_proj",
|
| 180 |
+
"51.self_attn.q_proj",
|
| 181 |
+
"41.self_attn.q_proj",
|
| 182 |
+
"54.self_attn.q_proj",
|
| 183 |
+
"language_model.layers.25.self_attn.v_proj",
|
| 184 |
+
"language_model.layers.6.self_attn.q_proj",
|
| 185 |
+
"language_model.layers.21.self_attn.v_proj",
|
| 186 |
+
"language_model.layers.0.self_attn.k_proj",
|
| 187 |
+
"48.self_attn.v_proj",
|
| 188 |
+
"57.self_attn.v_proj",
|
| 189 |
+
"31.self_attn.q_proj",
|
| 190 |
+
"35.self_attn.v_proj",
|
| 191 |
+
"52.self_attn.v_proj",
|
| 192 |
+
"49.self_attn.k_proj",
|
| 193 |
+
"language_model.layers.8.self_attn.q_proj",
|
| 194 |
+
"29.self_attn.k_proj",
|
| 195 |
+
"46.self_attn.v_proj",
|
| 196 |
+
"37.self_attn.v_proj",
|
| 197 |
+
"down_proj",
|
| 198 |
+
"up_proj",
|
| 199 |
+
"59.self_attn.v_proj",
|
| 200 |
+
"language_model.layers.1.self_attn.v_proj",
|
| 201 |
+
"56.self_attn.q_proj",
|
| 202 |
+
"language_model.layers.18.self_attn.k_proj",
|
| 203 |
+
"30.self_attn.k_proj",
|
| 204 |
+
"42.self_attn.q_proj",
|
| 205 |
+
"49.self_attn.v_proj",
|
| 206 |
+
"43.self_attn.q_proj",
|
| 207 |
+
"54.self_attn.k_proj",
|
| 208 |
+
"38.self_attn.q_proj",
|
| 209 |
+
"language_model.layers.14.self_attn.v_proj",
|
| 210 |
+
"46.self_attn.q_proj",
|
| 211 |
+
"43.self_attn.v_proj",
|
| 212 |
+
"49.self_attn.q_proj",
|
| 213 |
+
"language_model.layers.13.self_attn.v_proj",
|
| 214 |
+
"35.self_attn.k_proj",
|
| 215 |
+
"52.self_attn.q_proj",
|
| 216 |
+
"55.self_attn.q_proj",
|
| 217 |
+
"language_model.layers.1.self_attn.q_proj",
|
| 218 |
+
"61.self_attn.v_proj",
|
| 219 |
+
"55.self_attn.k_proj",
|
| 220 |
+
"43.self_attn.k_proj"
|
| 221 |
+
],
|
| 222 |
+
"target_parameters": null,
|
| 223 |
+
"task_type": null,
|
| 224 |
+
"trainable_token_indices": null,
|
| 225 |
+
"use_dora": false,
|
| 226 |
+
"use_qalora": false,
|
| 227 |
+
"use_rslora": false
|
| 228 |
+
}
|
gemma-3-27b-it/top-s0/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:78600827c4620494eb349138a4bd33fc43d3447affd834ec1a60be7a89a7515e
|
| 3 |
+
size 1816396048
|
gemma-3-27b-it/top-s0/config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "google/gemma-3-27b-it",
|
| 3 |
+
"data": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma/arms/TASKW-top-gemma.jsonl",
|
| 4 |
+
"run": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/gemma/runs/top-s0",
|
| 5 |
+
"steps": 0,
|
| 6 |
+
"batch": 16,
|
| 7 |
+
"micro": 1,
|
| 8 |
+
"seq_len": 4096,
|
| 9 |
+
"lr": 0.0001,
|
| 10 |
+
"warmup": 10,
|
| 11 |
+
"lora_rank": 64,
|
| 12 |
+
"lora_targets": "all-linear",
|
| 13 |
+
"lora_targets_file": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/code/gemma3_lora_targets.json",
|
| 14 |
+
"seed": 0,
|
| 15 |
+
"bos": true,
|
| 16 |
+
"no_merge": false,
|
| 17 |
+
"save_at": "",
|
| 18 |
+
"replay": "",
|
| 19 |
+
"replay_frac": 0.0,
|
| 20 |
+
"constant_lr": false,
|
| 21 |
+
"schedule": "hf-cosine",
|
| 22 |
+
"min_lr_frac": 0.0,
|
| 23 |
+
"warmup_ratio": 0.05,
|
| 24 |
+
"lora_dropout": 0.05,
|
| 25 |
+
"no_pack": true,
|
| 26 |
+
"chat_sft": true,
|
| 27 |
+
"data_frac": 1.0,
|
| 28 |
+
"base_adapter": "",
|
| 29 |
+
"ckpt_every": 25,
|
| 30 |
+
"kill_after": 0,
|
| 31 |
+
"method": "doc_lm_lora",
|
| 32 |
+
"world": 1
|
| 33 |
+
}
|
gemma-3-27b-it/top-s0/log.jsonl
ADDED
|
@@ -0,0 +1,143 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 1, "loss": 0.22505121769766853, "grad_norm": 0.21703773736953735, "lr": 2.857142857142857e-05, "epoch": 1, "t": 12.054490566253662}
|
| 2 |
+
{"step": 2, "loss": 0.24540961073642836, "grad_norm": 0.24778541922569275, "lr": 4.2857142857142856e-05, "epoch": 1, "t": 9.516986608505249}
|
| 3 |
+
{"step": 3, "loss": 0.22930878784614653, "grad_norm": 0.27400651574134827, "lr": 5.714285714285714e-05, "epoch": 1, "t": 10.105645656585693}
|
| 4 |
+
{"step": 4, "loss": 0.22194811920350035, "grad_norm": 0.23619204759597778, "lr": 7.142857142857143e-05, "epoch": 1, "t": 11.43257188796997}
|
| 5 |
+
{"step": 5, "loss": 0.22268119415085658, "grad_norm": 0.2712884545326233, "lr": 8.571428571428571e-05, "epoch": 1, "t": 9.65244460105896}
|
| 6 |
+
{"step": 6, "loss": 0.20631448412455675, "grad_norm": 0.2699175179004669, "lr": 0.0001, "epoch": 1, "t": 11.324448108673096}
|
| 7 |
+
{"step": 7, "loss": 0.24343506528702055, "grad_norm": 0.33858978748321533, "lr": 0.0001, "epoch": 1, "t": 9.84887146949768}
|
| 8 |
+
{"step": 8, "loss": 0.199659386745719, "grad_norm": 0.28170526027679443, "lr": 9.998666040558187e-05, "epoch": 1, "t": 9.421858787536621}
|
| 9 |
+
{"step": 9, "loss": 0.22586406636767833, "grad_norm": 0.27943772077560425, "lr": 9.994664874011863e-05, "epoch": 1, "t": 10.434593677520752}
|
| 10 |
+
{"step": 10, "loss": 0.22707384798576655, "grad_norm": 0.2697888910770416, "lr": 9.987998635318585e-05, "epoch": 1, "t": 10.50551176071167}
|
| 11 |
+
{"step": 11, "loss": 0.24545073671295564, "grad_norm": 0.2650054395198822, "lr": 9.978670881475172e-05, "epoch": 1, "t": 10.266358375549316}
|
| 12 |
+
{"step": 12, "loss": 0.19028063638530995, "grad_norm": 0.23152479529380798, "lr": 9.966686589619751e-05, "epoch": 1, "t": 10.103020191192627}
|
| 13 |
+
{"step": 13, "loss": 0.22196000356158432, "grad_norm": 1.3057767152786255, "lr": 9.952052154376026e-05, "epoch": 1, "t": 10.711141347885132}
|
| 14 |
+
{"step": 14, "loss": 0.21300532938160982, "grad_norm": 0.23710189759731293, "lr": 9.934775384441228e-05, "epoch": 1, "t": 10.228380918502808}
|
| 15 |
+
{"step": 15, "loss": 0.24879513348636922, "grad_norm": 0.2256975620985031, "lr": 9.91486549841951e-05, "epoch": 1, "t": 12.432562351226807}
|
| 16 |
+
{"step": 16, "loss": 0.20557237603502415, "grad_norm": 0.2202509343624115, "lr": 9.892333119903046e-05, "epoch": 1, "t": 11.608261823654175}
|
| 17 |
+
{"step": 17, "loss": 0.19515724440172577, "grad_norm": 0.23974862694740295, "lr": 9.867190271803465e-05, "epoch": 1, "t": 10.96469497680664}
|
| 18 |
+
{"step": 18, "loss": 0.1967107274436229, "grad_norm": 0.28855326771736145, "lr": 9.839450369936613e-05, "epoch": 1, "t": 8.185019731521606}
|
| 19 |
+
{"step": 19, "loss": 0.21986242121673952, "grad_norm": 0.2748332619667053, "lr": 9.809128215864097e-05, "epoch": 1, "t": 9.775784492492676}
|
| 20 |
+
{"step": 20, "loss": 0.21358585373547354, "grad_norm": 0.23560836911201477, "lr": 9.7762399889954e-05, "epoch": 1, "t": 9.42101788520813}
|
| 21 |
+
{"step": 21, "loss": 0.1925711565740602, "grad_norm": 0.24786853790283203, "lr": 9.74080323795483e-05, "epoch": 1, "t": 9.326472282409668}
|
| 22 |
+
{"step": 22, "loss": 0.238364584641796, "grad_norm": 0.24411338567733765, "lr": 9.702836871217839e-05, "epoch": 1, "t": 10.775306224822998}
|
| 23 |
+
{"step": 23, "loss": 0.1964091545150857, "grad_norm": 0.21883657574653625, "lr": 9.662361147021779e-05, "epoch": 1, "t": 11.269694805145264}
|
| 24 |
+
{"step": 24, "loss": 0.19321282507392137, "grad_norm": 0.2659012973308563, "lr": 9.619397662556435e-05, "epoch": 1, "t": 8.135856866836548}
|
| 25 |
+
{"step": 25, "loss": 0.24860771997723058, "grad_norm": 0.2588931918144226, "lr": 9.573969342440106e-05, "epoch": 1, "t": 9.904388666152954}
|
| 26 |
+
{"step": 26, "loss": 0.20770942510836127, "grad_norm": 0.2095607966184616, "lr": 9.526100426487409e-05, "epoch": 1, "t": 9.920934200286865}
|
| 27 |
+
{"step": 27, "loss": 0.2006321762915853, "grad_norm": 0.21268363296985626, "lr": 9.475816456775313e-05, "epoch": 1, "t": 10.83916163444519}
|
| 28 |
+
{"step": 28, "loss": 0.24594306693255452, "grad_norm": 0.22133034467697144, "lr": 9.423144264014277e-05, "epoch": 1, "t": 9.63451361656189}
|
| 29 |
+
{"step": 29, "loss": 0.19797796972210516, "grad_norm": 0.193564772605896, "lr": 9.368111953231848e-05, "epoch": 1, "t": 10.058306694030762}
|
| 30 |
+
{"step": 30, "loss": 0.20446276894851784, "grad_norm": 0.2121383547782898, "lr": 9.310748888776254e-05, "epoch": 1, "t": 9.457752704620361}
|
| 31 |
+
{"step": 31, "loss": 0.19927488339865662, "grad_norm": 0.19595593214035034, "lr": 9.251085678648072e-05, "epoch": 1, "t": 10.397753953933716}
|
| 32 |
+
{"step": 32, "loss": 0.21001055971556223, "grad_norm": 0.227900892496109, "lr": 9.189154158168292e-05, "epoch": 1, "t": 8.492723226547241}
|
| 33 |
+
{"step": 33, "loss": 0.225430331818045, "grad_norm": 0.21005074679851532, "lr": 9.124987372991511e-05, "epoch": 1, "t": 10.937589883804321}
|
| 34 |
+
{"step": 34, "loss": 0.20656682381318806, "grad_norm": 0.225301131606102, "lr": 9.058619561473307e-05, "epoch": 1, "t": 8.90029764175415}
|
| 35 |
+
{"step": 35, "loss": 0.22779996080064685, "grad_norm": 0.25957736372947693, "lr": 8.9900861364012e-05, "epoch": 1, "t": 9.490508079528809}
|
| 36 |
+
{"step": 36, "loss": 0.2126345234350297, "grad_norm": 0.23203442990779877, "lr": 8.919423666098969e-05, "epoch": 1, "t": 11.106785535812378}
|
| 37 |
+
{"step": 37, "loss": 0.2313187568391229, "grad_norm": 0.2481047362089157, "lr": 8.846669854914396e-05, "epoch": 1, "t": 9.578790187835693}
|
| 38 |
+
{"step": 38, "loss": 0.22694694870278267, "grad_norm": 0.22763553261756897, "lr": 8.77186352310082e-05, "epoch": 1, "t": 10.389273881912231}
|
| 39 |
+
{"step": 39, "loss": 0.19878314602618677, "grad_norm": 0.2309281826019287, "lr": 8.695044586103296e-05, "epoch": 1, "t": 10.274710416793823}
|
| 40 |
+
{"step": 40, "loss": 0.2167181975434751, "grad_norm": 0.24669983983039856, "lr": 8.616254033260351e-05, "epoch": 1, "t": 8.751072406768799}
|
| 41 |
+
{"step": 41, "loss": 0.24636918802136137, "grad_norm": 0.22230270504951477, "lr": 8.535533905932738e-05, "epoch": 1, "t": 11.74364447593689}
|
| 42 |
+
{"step": 42, "loss": 0.22521391857101258, "grad_norm": 0.21574485301971436, "lr": 8.452927275070858e-05, "epoch": 1, "t": 10.1707022190094}
|
| 43 |
+
{"step": 43, "loss": 0.1882047427302163, "grad_norm": 0.41809940338134766, "lr": 8.368478218232787e-05, "epoch": 1, "t": 10.103673219680786}
|
| 44 |
+
{"step": 44, "loss": 0.2175239191135965, "grad_norm": 0.2280159592628479, "lr": 8.282231796065214e-05, "epoch": 1, "t": 10.507831573486328}
|
| 45 |
+
{"step": 45, "loss": 0.22175029550468522, "grad_norm": 0.2500484883785248, "lr": 8.194234028259806e-05, "epoch": 1, "t": 9.814427375793457}
|
| 46 |
+
{"step": 46, "loss": 0.21215120287620454, "grad_norm": 0.19898876547813416, "lr": 8.104531868997858e-05, "epoch": 1, "t": 11.341955423355103}
|
| 47 |
+
{"step": 47, "loss": 0.18998032131318401, "grad_norm": 0.1909809708595276, "lr": 8.013173181896283e-05, "epoch": 1, "t": 10.59896445274353}
|
| 48 |
+
{"step": 48, "loss": 0.22170501866770564, "grad_norm": 0.20594868063926697, "lr": 7.920206714468383e-05, "epoch": 1, "t": 11.040373802185059}
|
| 49 |
+
{"step": 49, "loss": 0.215590952729174, "grad_norm": 0.22152985632419586, "lr": 7.82568207211296e-05, "epoch": 1, "t": 9.514512777328491}
|
| 50 |
+
{"step": 50, "loss": 0.2122216894412017, "grad_norm": 0.23917758464813232, "lr": 7.729649691645673e-05, "epoch": 1, "t": 9.262718200683594}
|
| 51 |
+
{"step": 51, "loss": 0.20459794676564677, "grad_norm": 0.19892755150794983, "lr": 7.63216081438678e-05, "epoch": 1, "t": 9.69749402999878}
|
| 52 |
+
{"step": 52, "loss": 0.21811848597818762, "grad_norm": 0.19602754712104797, "lr": 7.533267458819598e-05, "epoch": 1, "t": 10.440067768096924}
|
| 53 |
+
{"step": 53, "loss": 0.21435127394291206, "grad_norm": 0.23377782106399536, "lr": 7.433022392834282e-05, "epoch": 1, "t": 10.183382511138916}
|
| 54 |
+
{"step": 54, "loss": 0.20927664727682346, "grad_norm": 0.20169499516487122, "lr": 7.33147910557174e-05, "epoch": 1, "t": 11.860474109649658}
|
| 55 |
+
{"step": 55, "loss": 0.25015311167801246, "grad_norm": 0.22437871992588043, "lr": 7.228691778882693e-05, "epoch": 1, "t": 10.844437599182129}
|
| 56 |
+
{"step": 56, "loss": 0.2546206663716142, "grad_norm": 0.24023710191249847, "lr": 7.12471525841711e-05, "epoch": 1, "t": 11.428836822509766}
|
| 57 |
+
{"step": 57, "loss": 0.24053334362738954, "grad_norm": 0.25352856516838074, "lr": 7.019605024359474e-05, "epoch": 1, "t": 9.737306118011475}
|
| 58 |
+
{"step": 58, "loss": 0.22556048528626216, "grad_norm": 0.21189622581005096, "lr": 6.91341716182545e-05, "epoch": 1, "t": 12.7682044506073}
|
| 59 |
+
{"step": 59, "loss": 0.18967859314301058, "grad_norm": 0.22413399815559387, "lr": 6.806208330935766e-05, "epoch": 1, "t": 8.58992075920105}
|
| 60 |
+
{"step": 60, "loss": 0.21742234890089432, "grad_norm": 0.20742391049861908, "lr": 6.698035736583306e-05, "epoch": 1, "t": 11.425224542617798}
|
| 61 |
+
{"step": 61, "loss": 0.2135961299014334, "grad_norm": 0.2490430474281311, "lr": 6.588957097909508e-05, "epoch": 1, "t": 9.635200262069702}
|
| 62 |
+
{"step": 62, "loss": 0.21689573479258453, "grad_norm": 0.2328774482011795, "lr": 6.479030617506353e-05, "epoch": 1, "t": 9.334052324295044}
|
| 63 |
+
{"step": 63, "loss": 0.23668449976230663, "grad_norm": 0.21172454953193665, "lr": 6.368314950360415e-05, "epoch": 1, "t": 10.989005088806152}
|
| 64 |
+
{"step": 64, "loss": 0.2029248450160242, "grad_norm": 0.19610105454921722, "lr": 6.256869172555515e-05, "epoch": 1, "t": 10.582974672317505}
|
| 65 |
+
{"step": 65, "loss": 0.20569957869097744, "grad_norm": 0.206247016787529, "lr": 6.14475274975067e-05, "epoch": 1, "t": 11.197489738464355}
|
| 66 |
+
{"step": 66, "loss": 0.1983781010666692, "grad_norm": 0.19008895754814148, "lr": 6.0320255054501985e-05, "epoch": 1, "t": 11.482795476913452}
|
| 67 |
+
{"step": 67, "loss": 0.21081169469715408, "grad_norm": 0.1924007385969162, "lr": 5.918747589082853e-05, "epoch": 1, "t": 10.047632455825806}
|
| 68 |
+
{"step": 68, "loss": 0.22280451165881132, "grad_norm": 0.1961202621459961, "lr": 5.8049794439070646e-05, "epoch": 1, "t": 10.537443399429321}
|
| 69 |
+
{"step": 69, "loss": 0.20206253726379408, "grad_norm": 0.21997730433940887, "lr": 5.6907817747594116e-05, "epoch": 1, "t": 9.550058841705322}
|
| 70 |
+
{"step": 70, "loss": 0.2047861488675729, "grad_norm": 0.19869445264339447, "lr": 5.576215515663489e-05, "epoch": 1, "t": 9.825855016708374}
|
| 71 |
+
{"step": 71, "loss": 0.24573015028099182, "grad_norm": 0.26009121537208557, "lr": 5.4613417973165106e-05, "epoch": 1, "t": 8.839526891708374}
|
| 72 |
+
{"step": 72, "loss": 0.23537826049450813, "grad_norm": 0.18990512192249298, "lr": 5.346221914470959e-05, "epoch": 1, "t": 11.488405704498291}
|
| 73 |
+
{"step": 73, "loss": 0.2329046377068231, "grad_norm": 0.21941818296909332, "lr": 5.230917293228699e-05, "epoch": 1, "t": 9.61008334159851}
|
| 74 |
+
{"step": 74, "loss": 0.21540734000262854, "grad_norm": 0.20822052657604218, "lr": 5.115489458265006e-05, "epoch": 1, "t": 11.6629319190979}
|
| 75 |
+
{"step": 75, "loss": 0.24852643891305604, "grad_norm": 0.2189219444990158, "lr": 5e-05, "epoch": 1, "t": 11.586644411087036}
|
| 76 |
+
{"step": 76, "loss": 0.21666138455855163, "grad_norm": 0.17972682416439056, "lr": 4.884510541734995e-05, "epoch": 1, "t": 10.041048765182495}
|
| 77 |
+
{"step": 77, "loss": 0.19575391571305498, "grad_norm": 0.20232640206813812, "lr": 4.7690827067713035e-05, "epoch": 1, "t": 9.278618812561035}
|
| 78 |
+
{"step": 78, "loss": 0.2271962047940843, "grad_norm": 0.21232342720031738, "lr": 4.6537780855290426e-05, "epoch": 1, "t": 10.942893981933594}
|
| 79 |
+
{"step": 79, "loss": 0.23146169841220054, "grad_norm": 0.18582604825496674, "lr": 4.5386582026834906e-05, "epoch": 1, "t": 11.442293643951416}
|
| 80 |
+
{"step": 80, "loss": 0.22945719404400045, "grad_norm": 0.24335208535194397, "lr": 4.423784484336513e-05, "epoch": 1, "t": 8.328362226486206}
|
| 81 |
+
{"step": 81, "loss": 0.22532927693692714, "grad_norm": 0.19784903526306152, "lr": 4.30921822524059e-05, "epoch": 1, "t": 11.979598760604858}
|
| 82 |
+
{"step": 82, "loss": 0.219247266354893, "grad_norm": 0.22701801359653473, "lr": 4.195020556092935e-05, "epoch": 1, "t": 10.241004943847656}
|
| 83 |
+
{"step": 83, "loss": 0.24517080890883958, "grad_norm": 0.19996687769889832, "lr": 4.0812524109171476e-05, "epoch": 1, "t": 11.779272079467773}
|
| 84 |
+
{"step": 84, "loss": 0.20453113444419768, "grad_norm": 0.1847640722990036, "lr": 3.9679744945498026e-05, "epoch": 1, "t": 10.896613597869873}
|
| 85 |
+
{"step": 85, "loss": 0.1986247555989034, "grad_norm": 0.21294614672660828, "lr": 3.855247250249331e-05, "epoch": 1, "t": 11.634602546691895}
|
| 86 |
+
{"step": 86, "loss": 0.21482527971986304, "grad_norm": 0.20600295066833496, "lr": 3.7431308274444865e-05, "epoch": 1, "t": 11.46474313735962}
|
| 87 |
+
{"step": 87, "loss": 0.23929734182753465, "grad_norm": 0.20089924335479736, "lr": 3.631685049639586e-05, "epoch": 1, "t": 11.491093158721924}
|
| 88 |
+
{"step": 88, "loss": 0.22174171709610532, "grad_norm": 0.2268877625465393, "lr": 3.5209693824936484e-05, "epoch": 1, "t": 9.254255294799805}
|
| 89 |
+
{"step": 89, "loss": 0.2352444302937624, "grad_norm": 0.2403077632188797, "lr": 3.411042902090492e-05, "epoch": 1, "t": 8.510625123977661}
|
| 90 |
+
{"step": 90, "loss": 0.2030368671781914, "grad_norm": 0.2170497626066208, "lr": 3.301964263416693e-05, "epoch": 1, "t": 10.886615991592407}
|
| 91 |
+
{"step": 91, "loss": 0.2012726849903261, "grad_norm": 0.1920071244239807, "lr": 3.1937916690642356e-05, "epoch": 1, "t": 10.60518479347229}
|
| 92 |
+
{"step": 92, "loss": 0.21487654513840465, "grad_norm": 0.48220759630203247, "lr": 3.086582838174551e-05, "epoch": 1, "t": 11.4157874584198}
|
| 93 |
+
{"step": 93, "loss": 0.21617813801950178, "grad_norm": 0.1996578872203827, "lr": 2.980394975640526e-05, "epoch": 1, "t": 10.667667627334595}
|
| 94 |
+
{"step": 94, "loss": 0.213875618005616, "grad_norm": 0.21704868972301483, "lr": 2.875284741582892e-05, "epoch": 1, "t": 10.687480926513672}
|
| 95 |
+
{"step": 95, "loss": 0.20148991226217355, "grad_norm": 0.19102336466312408, "lr": 2.771308221117309e-05, "epoch": 1, "t": 8.994153499603271}
|
| 96 |
+
{"step": 96, "loss": 0.22516883546762628, "grad_norm": 0.2238171100616455, "lr": 2.668520894428259e-05, "epoch": 1, "t": 9.848058700561523}
|
| 97 |
+
{"step": 97, "loss": 0.23498747648507914, "grad_norm": 0.18566302955150604, "lr": 2.5669776071657192e-05, "epoch": 1, "t": 11.381712913513184}
|
| 98 |
+
{"step": 98, "loss": 0.20891795197826057, "grad_norm": 0.1896337866783142, "lr": 2.4667325411804036e-05, "epoch": 1, "t": 11.594501972198486}
|
| 99 |
+
{"step": 99, "loss": 0.20798248217693593, "grad_norm": 0.19009491801261902, "lr": 2.3678391856132204e-05, "epoch": 1, "t": 10.882423162460327}
|
| 100 |
+
{"step": 100, "loss": 0.22777974444781332, "grad_norm": 0.21916241943836212, "lr": 2.270350308354329e-05, "epoch": 1, "t": 8.92090129852295}
|
| 101 |
+
{"step": 101, "loss": 0.22950904547540296, "grad_norm": 0.19836682081222534, "lr": 2.1743179278870407e-05, "epoch": 1, "t": 12.38113021850586}
|
| 102 |
+
{"step": 102, "loss": 0.19720774813543393, "grad_norm": 0.18992555141448975, "lr": 2.079793285531618e-05, "epoch": 1, "t": 12.320561408996582}
|
| 103 |
+
{"step": 103, "loss": 0.21829094940529548, "grad_norm": 0.18572694063186646, "lr": 1.9868268181037185e-05, "epoch": 1, "t": 11.610471963882446}
|
| 104 |
+
{"step": 104, "loss": 0.2118741783090995, "grad_norm": 0.23200030624866486, "lr": 1.8954681310021433e-05, "epoch": 1, "t": 8.644611358642578}
|
| 105 |
+
{"step": 105, "loss": 0.20191033372910921, "grad_norm": 0.17914192378520966, "lr": 1.8057659717401947e-05, "epoch": 1, "t": 12.089706659317017}
|
| 106 |
+
{"step": 106, "loss": 0.25069651295919615, "grad_norm": 0.21586081385612488, "lr": 1.7177682039347876e-05, "epoch": 1, "t": 9.645750045776367}
|
| 107 |
+
{"step": 107, "loss": 0.21194703398225723, "grad_norm": 0.19013509154319763, "lr": 1.631521781767214e-05, "epoch": 1, "t": 11.319631814956665}
|
| 108 |
+
{"step": 108, "loss": 0.22893488565647194, "grad_norm": 0.21662981808185577, "lr": 1.5470727249291423e-05, "epoch": 1, "t": 10.19741153717041}
|
| 109 |
+
{"step": 109, "loss": 0.22705423582775164, "grad_norm": 0.23066486418247223, "lr": 1.4644660940672627e-05, "epoch": 1, "t": 8.92653489112854}
|
| 110 |
+
{"step": 110, "loss": 0.24917763492387654, "grad_norm": 0.23151402175426483, "lr": 1.3837459667396518e-05, "epoch": 1, "t": 10.738795757293701}
|
| 111 |
+
{"step": 111, "loss": 0.23403816625410656, "grad_norm": 0.20127059519290924, "lr": 1.3049554138967051e-05, "epoch": 1, "t": 11.840016841888428}
|
| 112 |
+
{"step": 112, "loss": 0.26001353187756193, "grad_norm": 0.246088445186615, "lr": 1.2281364768991804e-05, "epoch": 1, "t": 11.113420486450195}
|
| 113 |
+
{"step": 113, "loss": 0.2240116816353577, "grad_norm": 0.21575187146663666, "lr": 1.1533301450856054e-05, "epoch": 1, "t": 9.72246527671814}
|
| 114 |
+
{"step": 114, "loss": 0.22369254766458488, "grad_norm": 0.20920686423778534, "lr": 1.0805763339010327e-05, "epoch": 1, "t": 8.78674840927124}
|
| 115 |
+
{"step": 115, "loss": 0.20912365010812498, "grad_norm": 0.2666623890399933, "lr": 1.0099138635988026e-05, "epoch": 1, "t": 9.524937629699707}
|
| 116 |
+
{"step": 116, "loss": 0.20315464573109066, "grad_norm": 0.18049681186676025, "lr": 9.41380438526694e-06, "epoch": 1, "t": 12.498393774032593}
|
| 117 |
+
{"step": 117, "loss": 0.19453228550149712, "grad_norm": 0.2050487995147705, "lr": 8.75012627008489e-06, "epoch": 1, "t": 10.651840209960938}
|
| 118 |
+
{"step": 118, "loss": 0.20798705938511292, "grad_norm": 0.19289687275886536, "lr": 8.10845841831709e-06, "epoch": 1, "t": 10.539570093154907}
|
| 119 |
+
{"step": 119, "loss": 0.2331430976424099, "grad_norm": 0.21947957575321198, "lr": 7.489143213519301e-06, "epoch": 1, "t": 9.810571193695068}
|
| 120 |
+
{"step": 120, "loss": 0.23201207653560108, "grad_norm": 0.21218451857566833, "lr": 6.892511112237471e-06, "epoch": 1, "t": 10.715802907943726}
|
| 121 |
+
{"step": 121, "loss": 0.21074999272986514, "grad_norm": 0.21293194591999054, "lr": 6.318880467681526e-06, "epoch": 1, "t": 9.140785932540894}
|
| 122 |
+
{"step": 122, "loss": 0.21485237253398237, "grad_norm": 0.19227859377861023, "lr": 5.768557359857241e-06, "epoch": 1, "t": 11.459423542022705}
|
| 123 |
+
{"step": 123, "loss": 0.20440094586761795, "grad_norm": 0.23368176817893982, "lr": 5.241835432246889e-06, "epoch": 1, "t": 11.006624221801758}
|
| 124 |
+
{"step": 124, "loss": 0.21604196356349853, "grad_norm": 0.20674683153629303, "lr": 4.7389957351258955e-06, "epoch": 1, "t": 11.091724395751953}
|
| 125 |
+
{"step": 125, "loss": 0.1918921984028676, "grad_norm": 0.20515887439250946, "lr": 4.260306575598949e-06, "epoch": 1, "t": 10.054367780685425}
|
| 126 |
+
{"step": 126, "loss": 0.2155248782340382, "grad_norm": 0.17134223878383636, "lr": 3.8060233744356633e-06, "epoch": 1, "t": 12.724954843521118}
|
| 127 |
+
{"step": 127, "loss": 0.21280758940143343, "grad_norm": 0.20869047939777374, "lr": 3.376388529782215e-06, "epoch": 1, "t": 9.8471097946167}
|
| 128 |
+
{"step": 128, "loss": 0.20093681955694506, "grad_norm": 0.21226777136325836, "lr": 2.971631287821619e-06, "epoch": 1, "t": 9.71830940246582}
|
| 129 |
+
{"step": 129, "loss": 0.22660333895389922, "grad_norm": 0.19003553688526154, "lr": 2.591967620451707e-06, "epoch": 1, "t": 11.446911096572876}
|
| 130 |
+
{"step": 130, "loss": 0.21706106114979792, "grad_norm": 0.19154681265354156, "lr": 2.237600110046001e-06, "epoch": 1, "t": 11.401564121246338}
|
| 131 |
+
{"step": 131, "loss": 0.2186806542202629, "grad_norm": 0.21070224046707153, "lr": 1.908717841359048e-06, "epoch": 1, "t": 9.507974863052368}
|
| 132 |
+
{"step": 132, "loss": 0.25809210892890944, "grad_norm": 0.2355871945619583, "lr": 1.6054963006338742e-06, "epoch": 1, "t": 11.112389087677002}
|
| 133 |
+
{"step": 133, "loss": 0.21984846476706793, "grad_norm": 0.2116549015045166, "lr": 1.328097281965357e-06, "epoch": 1, "t": 10.80772614479065}
|
| 134 |
+
{"step": 134, "loss": 0.22923553089842794, "grad_norm": 0.2332264482975006, "lr": 1.0766688009695547e-06, "epoch": 1, "t": 10.506229400634766}
|
| 135 |
+
{"step": 135, "loss": 0.22959739099443716, "grad_norm": 0.2471969723701477, "lr": 8.513450158049108e-07, "epoch": 1, "t": 8.495505571365356}
|
| 136 |
+
{"step": 136, "loss": 0.2238175068600995, "grad_norm": 0.1862340122461319, "lr": 6.522461555877213e-07, "epoch": 1, "t": 10.767034530639648}
|
| 137 |
+
{"step": 137, "loss": 0.2077139842861446, "grad_norm": 0.19255129992961884, "lr": 4.794784562397458e-07, "epoch": 1, "t": 11.160823583602905}
|
| 138 |
+
{"step": 138, "loss": 0.20058147145811672, "grad_norm": 0.1884453445672989, "lr": 3.3313410380250154e-07, "epoch": 1, "t": 10.169746160507202}
|
| 139 |
+
{"step": 139, "loss": 0.2571327569272948, "grad_norm": 0.23495423793792725, "lr": 2.1329118524827662e-07, "epoch": 1, "t": 10.18721318244934}
|
| 140 |
+
{"step": 140, "loss": 0.24053145962129716, "grad_norm": 0.23312735557556152, "lr": 1.200136468141544e-07, "epoch": 1, "t": 9.572031736373901}
|
| 141 |
+
{"step": 141, "loss": 0.20576847730523853, "grad_norm": 0.19867339730262756, "lr": 5.3351259881379014e-08, "epoch": 1, "t": 9.011807680130005}
|
| 142 |
+
{"step": 142, "loss": 0.20119772438518577, "grad_norm": 0.2028876543045044, "lr": 1.3339594418138035e-08, "epoch": 1, "t": 11.02613377571106}
|
| 143 |
+
{"step": 143, "loss": 0.18990486103182108, "grad_norm": 0.21175725758075714, "lr": 0.0, "epoch": 1, "t": 11.162810802459717}
|
qwen3-14b/bottom-s0/README.md
ADDED
|
@@ -0,0 +1,206 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: /workspace-vast/jfonsecarivera/rlw-hf/hub/models--Qwen--Qwen3-14B/snapshots/40c069824f4251a91eefaf281ebe4c544efd3e18
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- base_model:adapter:/workspace-vast/jfonsecarivera/rlw-hf/hub/models--Qwen--Qwen3-14B/snapshots/40c069824f4251a91eefaf281ebe4c544efd3e18
|
| 6 |
+
- lora
|
| 7 |
+
- transformers
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# Model Card for Model ID
|
| 11 |
+
|
| 12 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
## Model Details
|
| 17 |
+
|
| 18 |
+
### Model Description
|
| 19 |
+
|
| 20 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
- **Developed by:** [More Information Needed]
|
| 25 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 26 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 27 |
+
- **Model type:** [More Information Needed]
|
| 28 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 29 |
+
- **License:** [More Information Needed]
|
| 30 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 31 |
+
|
| 32 |
+
### Model Sources [optional]
|
| 33 |
+
|
| 34 |
+
<!-- Provide the basic links for the model. -->
|
| 35 |
+
|
| 36 |
+
- **Repository:** [More Information Needed]
|
| 37 |
+
- **Paper [optional]:** [More Information Needed]
|
| 38 |
+
- **Demo [optional]:** [More Information Needed]
|
| 39 |
+
|
| 40 |
+
## Uses
|
| 41 |
+
|
| 42 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 43 |
+
|
| 44 |
+
### Direct Use
|
| 45 |
+
|
| 46 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 47 |
+
|
| 48 |
+
[More Information Needed]
|
| 49 |
+
|
| 50 |
+
### Downstream Use [optional]
|
| 51 |
+
|
| 52 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 53 |
+
|
| 54 |
+
[More Information Needed]
|
| 55 |
+
|
| 56 |
+
### Out-of-Scope Use
|
| 57 |
+
|
| 58 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 59 |
+
|
| 60 |
+
[More Information Needed]
|
| 61 |
+
|
| 62 |
+
## Bias, Risks, and Limitations
|
| 63 |
+
|
| 64 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 65 |
+
|
| 66 |
+
[More Information Needed]
|
| 67 |
+
|
| 68 |
+
### Recommendations
|
| 69 |
+
|
| 70 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 71 |
+
|
| 72 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 73 |
+
|
| 74 |
+
## How to Get Started with the Model
|
| 75 |
+
|
| 76 |
+
Use the code below to get started with the model.
|
| 77 |
+
|
| 78 |
+
[More Information Needed]
|
| 79 |
+
|
| 80 |
+
## Training Details
|
| 81 |
+
|
| 82 |
+
### Training Data
|
| 83 |
+
|
| 84 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 85 |
+
|
| 86 |
+
[More Information Needed]
|
| 87 |
+
|
| 88 |
+
### Training Procedure
|
| 89 |
+
|
| 90 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 91 |
+
|
| 92 |
+
#### Preprocessing [optional]
|
| 93 |
+
|
| 94 |
+
[More Information Needed]
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
#### Training Hyperparameters
|
| 98 |
+
|
| 99 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 100 |
+
|
| 101 |
+
#### Speeds, Sizes, Times [optional]
|
| 102 |
+
|
| 103 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 104 |
+
|
| 105 |
+
[More Information Needed]
|
| 106 |
+
|
| 107 |
+
## Evaluation
|
| 108 |
+
|
| 109 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 110 |
+
|
| 111 |
+
### Testing Data, Factors & Metrics
|
| 112 |
+
|
| 113 |
+
#### Testing Data
|
| 114 |
+
|
| 115 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 116 |
+
|
| 117 |
+
[More Information Needed]
|
| 118 |
+
|
| 119 |
+
#### Factors
|
| 120 |
+
|
| 121 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 122 |
+
|
| 123 |
+
[More Information Needed]
|
| 124 |
+
|
| 125 |
+
#### Metrics
|
| 126 |
+
|
| 127 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 128 |
+
|
| 129 |
+
[More Information Needed]
|
| 130 |
+
|
| 131 |
+
### Results
|
| 132 |
+
|
| 133 |
+
[More Information Needed]
|
| 134 |
+
|
| 135 |
+
#### Summary
|
| 136 |
+
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
## Model Examination [optional]
|
| 140 |
+
|
| 141 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 142 |
+
|
| 143 |
+
[More Information Needed]
|
| 144 |
+
|
| 145 |
+
## Environmental Impact
|
| 146 |
+
|
| 147 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 148 |
+
|
| 149 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 150 |
+
|
| 151 |
+
- **Hardware Type:** [More Information Needed]
|
| 152 |
+
- **Hours used:** [More Information Needed]
|
| 153 |
+
- **Cloud Provider:** [More Information Needed]
|
| 154 |
+
- **Compute Region:** [More Information Needed]
|
| 155 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 156 |
+
|
| 157 |
+
## Technical Specifications [optional]
|
| 158 |
+
|
| 159 |
+
### Model Architecture and Objective
|
| 160 |
+
|
| 161 |
+
[More Information Needed]
|
| 162 |
+
|
| 163 |
+
### Compute Infrastructure
|
| 164 |
+
|
| 165 |
+
[More Information Needed]
|
| 166 |
+
|
| 167 |
+
#### Hardware
|
| 168 |
+
|
| 169 |
+
[More Information Needed]
|
| 170 |
+
|
| 171 |
+
#### Software
|
| 172 |
+
|
| 173 |
+
[More Information Needed]
|
| 174 |
+
|
| 175 |
+
## Citation [optional]
|
| 176 |
+
|
| 177 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 178 |
+
|
| 179 |
+
**BibTeX:**
|
| 180 |
+
|
| 181 |
+
[More Information Needed]
|
| 182 |
+
|
| 183 |
+
**APA:**
|
| 184 |
+
|
| 185 |
+
[More Information Needed]
|
| 186 |
+
|
| 187 |
+
## Glossary [optional]
|
| 188 |
+
|
| 189 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 190 |
+
|
| 191 |
+
[More Information Needed]
|
| 192 |
+
|
| 193 |
+
## More Information [optional]
|
| 194 |
+
|
| 195 |
+
[More Information Needed]
|
| 196 |
+
|
| 197 |
+
## Model Card Authors [optional]
|
| 198 |
+
|
| 199 |
+
[More Information Needed]
|
| 200 |
+
|
| 201 |
+
## Model Card Contact
|
| 202 |
+
|
| 203 |
+
[More Information Needed]
|
| 204 |
+
### Framework versions
|
| 205 |
+
|
| 206 |
+
- PEFT 0.17.1
|
qwen3-14b/bottom-s0/adapter_config.json
ADDED
|
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": {
|
| 4 |
+
"base_model_class": "Qwen3ForCausalLM",
|
| 5 |
+
"parent_library": "transformers.models.qwen3.modeling_qwen3"
|
| 6 |
+
},
|
| 7 |
+
"base_model_name_or_path": "/workspace-vast/jfonsecarivera/rlw-hf/hub/models--Qwen--Qwen3-14B/snapshots/40c069824f4251a91eefaf281ebe4c544efd3e18",
|
| 8 |
+
"bias": "none",
|
| 9 |
+
"corda_config": null,
|
| 10 |
+
"eva_config": null,
|
| 11 |
+
"exclude_modules": null,
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": true,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layer_replication": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"layers_to_transform": null,
|
| 18 |
+
"loftq_config": {},
|
| 19 |
+
"lora_alpha": 128,
|
| 20 |
+
"lora_bias": false,
|
| 21 |
+
"lora_dropout": 0.05,
|
| 22 |
+
"megatron_config": null,
|
| 23 |
+
"megatron_core": "megatron.core",
|
| 24 |
+
"modules_to_save": null,
|
| 25 |
+
"peft_type": "LORA",
|
| 26 |
+
"qalora_group_size": 16,
|
| 27 |
+
"r": 64,
|
| 28 |
+
"rank_pattern": {},
|
| 29 |
+
"revision": null,
|
| 30 |
+
"target_modules": [
|
| 31 |
+
"o_proj",
|
| 32 |
+
"up_proj",
|
| 33 |
+
"gate_proj",
|
| 34 |
+
"q_proj",
|
| 35 |
+
"down_proj",
|
| 36 |
+
"k_proj",
|
| 37 |
+
"v_proj"
|
| 38 |
+
],
|
| 39 |
+
"target_parameters": null,
|
| 40 |
+
"task_type": null,
|
| 41 |
+
"trainable_token_indices": null,
|
| 42 |
+
"use_dora": false,
|
| 43 |
+
"use_qalora": false,
|
| 44 |
+
"use_rslora": false
|
| 45 |
+
}
|
qwen3-14b/bottom-s0/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8106cd1534450e97a758c59f6978f271c3d711bd1dfc9e67d2ded0bd31577a9e
|
| 3 |
+
size 1027680256
|
qwen3-14b/bottom-s0/config.json
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "/workspace-vast/jfonsecarivera/rlw-hf/hub/models--Qwen--Qwen3-14B/snapshots/40c069824f4251a91eefaf281ebe4c544efd3e18",
|
| 3 |
+
"data": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/qwen/arms/TASKW-bottom-qwen.jsonl",
|
| 4 |
+
"run": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/qwen/runs/bottom-s0",
|
| 5 |
+
"steps": 0,
|
| 6 |
+
"batch": 16,
|
| 7 |
+
"micro": 1,
|
| 8 |
+
"seq_len": 4096,
|
| 9 |
+
"lr": 0.0001,
|
| 10 |
+
"warmup": 10,
|
| 11 |
+
"lora_rank": 64,
|
| 12 |
+
"lora_targets": "all-linear",
|
| 13 |
+
"lora_targets_file": "",
|
| 14 |
+
"seed": 0,
|
| 15 |
+
"bos": false,
|
| 16 |
+
"no_merge": false,
|
| 17 |
+
"save_at": "",
|
| 18 |
+
"replay": "",
|
| 19 |
+
"replay_frac": 0.0,
|
| 20 |
+
"constant_lr": false,
|
| 21 |
+
"schedule": "hf-cosine",
|
| 22 |
+
"min_lr_frac": 0.0,
|
| 23 |
+
"warmup_ratio": 0.05,
|
| 24 |
+
"lora_dropout": 0.05,
|
| 25 |
+
"no_pack": true,
|
| 26 |
+
"chat_sft": true,
|
| 27 |
+
"data_frac": 1.0,
|
| 28 |
+
"base_adapter": "",
|
| 29 |
+
"ckpt_every": 25,
|
| 30 |
+
"kill_after": 0,
|
| 31 |
+
"train_token_rows": "",
|
| 32 |
+
"method": "doc_lm_lora",
|
| 33 |
+
"world": 1
|
| 34 |
+
}
|
qwen3-14b/bottom-s0/log.jsonl
ADDED
|
@@ -0,0 +1,296 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 1, "loss": 0.1752485499130772, "grad_norm": 0.2273269146680832, "lr": 1.3333333333333333e-05, "epoch": 1, "t": 7.2622599601745605}
|
| 2 |
+
{"step": 2, "loss": 0.16197575131929842, "grad_norm": 0.18963158130645752, "lr": 2e-05, "epoch": 1, "t": 6.842254877090454}
|
| 3 |
+
{"step": 3, "loss": 0.14372770245109295, "grad_norm": 0.2149733006954193, "lr": 2.6666666666666667e-05, "epoch": 1, "t": 6.149490594863892}
|
| 4 |
+
{"step": 4, "loss": 0.17578186904840032, "grad_norm": 0.2089262306690216, "lr": 3.3333333333333335e-05, "epoch": 1, "t": 6.206242561340332}
|
| 5 |
+
{"step": 5, "loss": 0.16244048400005653, "grad_norm": 0.20248648524284363, "lr": 4e-05, "epoch": 1, "t": 6.000253200531006}
|
| 6 |
+
{"step": 6, "loss": 0.14900893488448583, "grad_norm": 0.20379552245140076, "lr": 4.666666666666667e-05, "epoch": 1, "t": 6.036116123199463}
|
| 7 |
+
{"step": 7, "loss": 0.12102704053220495, "grad_norm": 0.25668591260910034, "lr": 5.333333333333333e-05, "epoch": 1, "t": 6.182107210159302}
|
| 8 |
+
{"step": 8, "loss": 0.13636366648148307, "grad_norm": 0.5128408074378967, "lr": 6e-05, "epoch": 1, "t": 5.9150471687316895}
|
| 9 |
+
{"step": 9, "loss": 0.16004198848871895, "grad_norm": 0.26240718364715576, "lr": 6.666666666666667e-05, "epoch": 1, "t": 5.8975677490234375}
|
| 10 |
+
{"step": 10, "loss": 0.13735793998360057, "grad_norm": 0.3030981719493866, "lr": 7.333333333333333e-05, "epoch": 1, "t": 5.992259740829468}
|
| 11 |
+
{"step": 11, "loss": 0.12292885473227813, "grad_norm": 0.2684709131717682, "lr": 8e-05, "epoch": 1, "t": 6.029702663421631}
|
| 12 |
+
{"step": 12, "loss": 0.1540826850778417, "grad_norm": 0.24784551560878754, "lr": 8.666666666666667e-05, "epoch": 1, "t": 5.965942621231079}
|
| 13 |
+
{"step": 13, "loss": 0.125180684584017, "grad_norm": 0.21856532990932465, "lr": 9.333333333333334e-05, "epoch": 1, "t": 5.84442925453186}
|
| 14 |
+
{"step": 14, "loss": 0.1336468793688672, "grad_norm": 0.1753084808588028, "lr": 0.0001, "epoch": 1, "t": 6.03787899017334}
|
| 15 |
+
{"step": 15, "loss": 0.11304400092002005, "grad_norm": 0.23694230616092682, "lr": 0.0001, "epoch": 1, "t": 5.878647565841675}
|
| 16 |
+
{"step": 16, "loss": 0.11225881042186095, "grad_norm": 0.2033313512802124, "lr": 9.999687519737639e-05, "epoch": 1, "t": 5.923731327056885}
|
| 17 |
+
{"step": 17, "loss": 0.11136573642254222, "grad_norm": 0.16831859946250916, "lr": 9.998750118008115e-05, "epoch": 1, "t": 5.944855690002441}
|
| 18 |
+
{"step": 18, "loss": 0.1436524074572562, "grad_norm": 0.22492867708206177, "lr": 9.997187911979252e-05, "epoch": 1, "t": 4.660308122634888}
|
| 19 |
+
{"step": 19, "loss": 0.10213176040634413, "grad_norm": 0.19424419105052948, "lr": 9.995001096914461e-05, "epoch": 1, "t": 4.2838804721832275}
|
| 20 |
+
{"step": 20, "loss": 0.1249386400594415, "grad_norm": 0.17311154305934906, "lr": 9.992189946148366e-05, "epoch": 1, "t": 5.620733022689819}
|
| 21 |
+
{"step": 21, "loss": 0.11955899123787346, "grad_norm": 0.16216126084327698, "lr": 9.988754811052616e-05, "epoch": 1, "t": 6.133888244628906}
|
| 22 |
+
{"step": 22, "loss": 0.1367188526761167, "grad_norm": 0.1654491275548935, "lr": 9.984696120991978e-05, "epoch": 1, "t": 6.002611398696899}
|
| 23 |
+
{"step": 23, "loss": 0.10915484067420786, "grad_norm": 0.13785584270954132, "lr": 9.980014383270668e-05, "epoch": 1, "t": 5.492818593978882}
|
| 24 |
+
{"step": 24, "loss": 0.11355082914945644, "grad_norm": 0.16870281100273132, "lr": 9.974710183068935e-05, "epoch": 1, "t": 5.871350049972534}
|
| 25 |
+
{"step": 25, "loss": 0.10453763878630122, "grad_norm": 0.19669336080551147, "lr": 9.968784183369929e-05, "epoch": 1, "t": 5.564558267593384}
|
| 26 |
+
{"step": 26, "loss": 0.09042725515465194, "grad_norm": 0.1999388039112091, "lr": 9.962237124876828e-05, "epoch": 1, "t": 5.271064758300781}
|
| 27 |
+
{"step": 27, "loss": 0.14823067171213114, "grad_norm": 0.184713676571846, "lr": 9.955069825920249e-05, "epoch": 1, "t": 5.541146755218506}
|
| 28 |
+
{"step": 28, "loss": 0.11725434708859521, "grad_norm": 0.19592486321926117, "lr": 9.947283182355982e-05, "epoch": 1, "t": 5.536479234695435}
|
| 29 |
+
{"step": 29, "loss": 0.12784560044026239, "grad_norm": 0.19796410202980042, "lr": 9.938878167452992e-05, "epoch": 1, "t": 5.433374881744385}
|
| 30 |
+
{"step": 30, "loss": 0.13446450256655165, "grad_norm": 0.1550251692533493, "lr": 9.929855831771786e-05, "epoch": 1, "t": 5.4044764041900635}
|
| 31 |
+
{"step": 31, "loss": 0.12071090125751266, "grad_norm": 0.17487181723117828, "lr": 9.92021730303309e-05, "epoch": 1, "t": 5.400969743728638}
|
| 32 |
+
{"step": 32, "loss": 0.09524619597507142, "grad_norm": 0.193100705742836, "lr": 9.909963785976903e-05, "epoch": 1, "t": 5.389806509017944}
|
| 33 |
+
{"step": 33, "loss": 0.12547221358709215, "grad_norm": 0.158849835395813, "lr": 9.899096562211902e-05, "epoch": 1, "t": 5.163612127304077}
|
| 34 |
+
{"step": 34, "loss": 0.14120039416162725, "grad_norm": 0.16312792897224426, "lr": 9.887616990055262e-05, "epoch": 1, "t": 4.6968300342559814}
|
| 35 |
+
{"step": 35, "loss": 0.11725060856240427, "grad_norm": 0.16907823085784912, "lr": 9.875526504362869e-05, "epoch": 1, "t": 4.679459810256958}
|
| 36 |
+
{"step": 36, "loss": 0.11519571962900538, "grad_norm": 0.39077630639076233, "lr": 9.86282661634998e-05, "epoch": 1, "t": 5.017693042755127}
|
| 37 |
+
{"step": 37, "loss": 0.10394156329469015, "grad_norm": 0.19277916848659515, "lr": 9.849518913402334e-05, "epoch": 1, "t": 4.630793809890747}
|
| 38 |
+
{"step": 38, "loss": 0.13825914087408372, "grad_norm": 0.1808989942073822, "lr": 9.835605058877729e-05, "epoch": 1, "t": 4.734872341156006}
|
| 39 |
+
{"step": 39, "loss": 0.14669809025035924, "grad_norm": 0.15419261157512665, "lr": 9.821086791898134e-05, "epoch": 1, "t": 5.179617643356323}
|
| 40 |
+
{"step": 40, "loss": 0.1154066498297602, "grad_norm": 0.19307547807693481, "lr": 9.805965927132295e-05, "epoch": 1, "t": 4.3937201499938965}
|
| 41 |
+
{"step": 41, "loss": 0.15095672952380426, "grad_norm": 0.19648480415344238, "lr": 9.79024435456893e-05, "epoch": 1, "t": 3.8020637035369873}
|
| 42 |
+
{"step": 42, "loss": 0.1332015265766372, "grad_norm": 0.16474975645542145, "lr": 9.773924039280487e-05, "epoch": 1, "t": 4.042931318283081}
|
| 43 |
+
{"step": 43, "loss": 0.1271522599560576, "grad_norm": 0.16134263575077057, "lr": 9.75700702117753e-05, "epoch": 1, "t": 4.079339265823364}
|
| 44 |
+
{"step": 44, "loss": 0.10394006987216055, "grad_norm": 0.17610692977905273, "lr": 9.739495414753753e-05, "epoch": 1, "t": 3.747913360595703}
|
| 45 |
+
{"step": 45, "loss": 0.1147655960400625, "grad_norm": 0.14709608256816864, "lr": 9.721391408821711e-05, "epoch": 1, "t": 3.8322336673736572}
|
| 46 |
+
{"step": 46, "loss": 0.1092554634260399, "grad_norm": 0.16116903722286224, "lr": 9.702697266239212e-05, "epoch": 1, "t": 3.805285930633545}
|
| 47 |
+
{"step": 47, "loss": 0.13763013517263303, "grad_norm": 0.1721290647983551, "lr": 9.683415323626485e-05, "epoch": 1, "t": 3.5704293251037598}
|
| 48 |
+
{"step": 48, "loss": 0.09924356610977464, "grad_norm": 0.1799483746290207, "lr": 9.663547991074127e-05, "epoch": 1, "t": 3.1370041370391846}
|
| 49 |
+
{"step": 49, "loss": 0.12897537376202683, "grad_norm": 0.17098456621170044, "lr": 9.643097751841854e-05, "epoch": 1, "t": 3.6654934883117676}
|
| 50 |
+
{"step": 50, "loss": 0.11030346398450891, "grad_norm": 0.1793755441904068, "lr": 9.622067162048112e-05, "epoch": 1, "t": 3.2133631706237793}
|
| 51 |
+
{"step": 51, "loss": 0.10788446882157156, "grad_norm": 0.1423107087612152, "lr": 9.600458850350588e-05, "epoch": 1, "t": 3.9863667488098145}
|
| 52 |
+
{"step": 52, "loss": 0.12193814227010509, "grad_norm": 0.27060258388519287, "lr": 9.578275517617645e-05, "epoch": 1, "t": 3.823734998703003}
|
| 53 |
+
{"step": 53, "loss": 0.13209376636596667, "grad_norm": 0.14064420759677887, "lr": 9.555519936590738e-05, "epoch": 1, "t": 4.1670823097229}
|
| 54 |
+
{"step": 54, "loss": 0.10726402326110894, "grad_norm": 0.12126035988330841, "lr": 9.532194951537838e-05, "epoch": 1, "t": 4.2917704582214355}
|
| 55 |
+
{"step": 55, "loss": 0.13402982711183978, "grad_norm": 0.16427747905254364, "lr": 9.508303477897924e-05, "epoch": 1, "t": 3.693385124206543}
|
| 56 |
+
{"step": 56, "loss": 0.11277710025861432, "grad_norm": 0.21456760168075562, "lr": 9.483848501916578e-05, "epoch": 1, "t": 3.2162392139434814}
|
| 57 |
+
{"step": 57, "loss": 0.09820183289992059, "grad_norm": 0.18677082657814026, "lr": 9.458833080272722e-05, "epoch": 1, "t": 3.2174220085144043}
|
| 58 |
+
{"step": 58, "loss": 0.0888994905341741, "grad_norm": 0.14535124599933624, "lr": 9.433260339696563e-05, "epoch": 1, "t": 3.962379217147827}
|
| 59 |
+
{"step": 59, "loss": 0.13740586502302582, "grad_norm": 0.17558394372463226, "lr": 9.407133476578778e-05, "epoch": 1, "t": 4.730549573898315}
|
| 60 |
+
{"step": 60, "loss": 0.12455208271865528, "grad_norm": 0.14261916279792786, "lr": 9.38045575657098e-05, "epoch": 1, "t": 4.914440870285034}
|
| 61 |
+
{"step": 61, "loss": 0.11913379427388204, "grad_norm": 0.1595505177974701, "lr": 9.353230514177552e-05, "epoch": 1, "t": 4.58712100982666}
|
| 62 |
+
{"step": 62, "loss": 0.1598939223347861, "grad_norm": 0.16470298171043396, "lr": 9.325461152338846e-05, "epoch": 1, "t": 4.833375930786133}
|
| 63 |
+
{"step": 63, "loss": 0.10565623049247465, "grad_norm": 0.14636284112930298, "lr": 9.297151142005851e-05, "epoch": 1, "t": 4.798422336578369}
|
| 64 |
+
{"step": 64, "loss": 0.12876437884645323, "grad_norm": 0.15872706472873688, "lr": 9.268304021706349e-05, "epoch": 1, "t": 4.652063608169556}
|
| 65 |
+
{"step": 65, "loss": 0.12992243309997886, "grad_norm": 0.1705460548400879, "lr": 9.23892339710263e-05, "epoch": 1, "t": 5.494608640670776}
|
| 66 |
+
{"step": 66, "loss": 0.12887709248450493, "grad_norm": 0.1533844918012619, "lr": 9.209012940540805e-05, "epoch": 1, "t": 4.818042993545532}
|
| 67 |
+
{"step": 67, "loss": 0.09908132047069315, "grad_norm": 0.16532683372497559, "lr": 9.178576390591802e-05, "epoch": 1, "t": 4.478140592575073}
|
| 68 |
+
{"step": 68, "loss": 0.1082844658700641, "grad_norm": 0.16935817897319794, "lr": 9.147617551584066e-05, "epoch": 1, "t": 4.695854663848877}
|
| 69 |
+
{"step": 69, "loss": 0.09549259292030872, "grad_norm": 0.14777007699012756, "lr": 9.116140293128051e-05, "epoch": 1, "t": 3.5415735244750977}
|
| 70 |
+
{"step": 70, "loss": 0.14593881517626658, "grad_norm": 0.16458331048488617, "lr": 9.084148549632547e-05, "epoch": 1, "t": 3.7696597576141357}
|
| 71 |
+
{"step": 71, "loss": 0.15460061061684927, "grad_norm": 0.1709330976009369, "lr": 9.051646319812918e-05, "epoch": 1, "t": 4.229056358337402}
|
| 72 |
+
{"step": 72, "loss": 0.09152270898436607, "grad_norm": 0.14009197056293488, "lr": 9.018637666191283e-05, "epoch": 1, "t": 4.38850998878479}
|
| 73 |
+
{"step": 73, "loss": 0.12666737965595098, "grad_norm": 0.15965832769870758, "lr": 8.985126714588738e-05, "epoch": 1, "t": 3.629696846008301}
|
| 74 |
+
{"step": 74, "loss": 0.09974380793505336, "grad_norm": 0.1462961733341217, "lr": 8.951117653609666e-05, "epoch": 1, "t": 4.481092214584351}
|
| 75 |
+
{"step": 75, "loss": 0.11864136477154336, "grad_norm": 0.14830192923545837, "lr": 8.916614734118184e-05, "epoch": 1, "t": 5.081501722335815}
|
| 76 |
+
{"step": 76, "loss": 0.14721758977480684, "grad_norm": 0.13556744158267975, "lr": 8.881622268706825e-05, "epoch": 1, "t": 5.222254276275635}
|
| 77 |
+
{"step": 77, "loss": 0.10595594941308704, "grad_norm": 0.1493482142686844, "lr": 8.8461446311575e-05, "epoch": 1, "t": 5.006181716918945}
|
| 78 |
+
{"step": 78, "loss": 0.1284816624444724, "grad_norm": 0.15887455642223358, "lr": 8.810186255894803e-05, "epoch": 1, "t": 5.144675254821777}
|
| 79 |
+
{"step": 79, "loss": 0.10622403047380655, "grad_norm": 0.1342378407716751, "lr": 8.773751637431748e-05, "epoch": 1, "t": 5.2127625942230225}
|
| 80 |
+
{"step": 80, "loss": 0.11771947886193393, "grad_norm": 0.14702412486076355, "lr": 8.736845329807993e-05, "epoch": 1, "t": 5.077196359634399}
|
| 81 |
+
{"step": 81, "loss": 0.08715652897214649, "grad_norm": 0.14258985221385956, "lr": 8.69947194602061e-05, "epoch": 1, "t": 5.166703939437866}
|
| 82 |
+
{"step": 82, "loss": 0.1407249934787319, "grad_norm": 0.1574077159166336, "lr": 8.66163615744751e-05, "epoch": 1, "t": 4.982941389083862}
|
| 83 |
+
{"step": 83, "loss": 0.10946292864949082, "grad_norm": 0.1504049152135849, "lr": 8.623342693263548e-05, "epoch": 1, "t": 4.9891884326934814}
|
| 84 |
+
{"step": 84, "loss": 0.11453432910343306, "grad_norm": 0.14204095304012299, "lr": 8.584596339849417e-05, "epoch": 1, "t": 4.742063760757446}
|
| 85 |
+
{"step": 85, "loss": 0.1411642906505004, "grad_norm": 0.14732734858989716, "lr": 8.545401940193392e-05, "epoch": 1, "t": 4.421473026275635}
|
| 86 |
+
{"step": 86, "loss": 0.12692128487083257, "grad_norm": 0.1760338395833969, "lr": 8.505764393285984e-05, "epoch": 1, "t": 3.8381481170654297}
|
| 87 |
+
{"step": 87, "loss": 0.1448906666822031, "grad_norm": 0.14406467974185944, "lr": 8.46568865350762e-05, "epoch": 1, "t": 4.42296838760376}
|
| 88 |
+
{"step": 88, "loss": 0.09271264168849525, "grad_norm": 0.13818959891796112, "lr": 8.425179730009368e-05, "epoch": 1, "t": 4.221201181411743}
|
| 89 |
+
{"step": 89, "loss": 0.09044018554625603, "grad_norm": 0.1389867067337036, "lr": 8.384242686086848e-05, "epoch": 1, "t": 4.587378978729248}
|
| 90 |
+
{"step": 90, "loss": 0.09728664608288758, "grad_norm": 0.14242956042289734, "lr": 8.342882638547351e-05, "epoch": 1, "t": 4.411623239517212}
|
| 91 |
+
{"step": 91, "loss": 0.1044411832391533, "grad_norm": 0.16613753139972687, "lr": 8.301104757070274e-05, "epoch": 1, "t": 3.752352714538574}
|
| 92 |
+
{"step": 92, "loss": 0.12271639306552244, "grad_norm": 0.1773744821548462, "lr": 8.258914263560971e-05, "epoch": 1, "t": 3.892817258834839}
|
| 93 |
+
{"step": 93, "loss": 0.1380967544534043, "grad_norm": 0.15400643646717072, "lr": 8.216316431498028e-05, "epoch": 1, "t": 4.50601053237915}
|
| 94 |
+
{"step": 94, "loss": 0.1510967011114474, "grad_norm": 0.13552269339561462, "lr": 8.173316585274145e-05, "epoch": 1, "t": 4.505499601364136}
|
| 95 |
+
{"step": 95, "loss": 0.09745585613861579, "grad_norm": 0.15307575464248657, "lr": 8.129920099530607e-05, "epoch": 1, "t": 3.8790650367736816}
|
| 96 |
+
{"step": 96, "loss": 0.12510263738168714, "grad_norm": 0.1437438428401947, "lr": 8.086132398485524e-05, "epoch": 1, "t": 4.18375563621521}
|
| 97 |
+
{"step": 97, "loss": 0.11138657050017126, "grad_norm": 0.137197345495224, "lr": 8.041958955255814e-05, "epoch": 1, "t": 4.412342071533203}
|
| 98 |
+
{"step": 98, "loss": 0.1041658407786736, "grad_norm": 0.1670880913734436, "lr": 7.99740529117313e-05, "epoch": 1, "t": 3.70011043548584}
|
| 99 |
+
{"step": 99, "loss": 0.1297504555360572, "grad_norm": 0.152018204331398, "lr": 7.952476975093729e-05, "epoch": 1, "t": 4.8253562450408936}
|
| 100 |
+
{"step": 100, "loss": 0.15818131143754804, "grad_norm": 0.17151501774787903, "lr": 7.907179622702408e-05, "epoch": 1, "t": 5.031487941741943}
|
| 101 |
+
{"step": 101, "loss": 0.1375515077137357, "grad_norm": 0.14399169385433197, "lr": 7.861518895810596e-05, "epoch": 1, "t": 3.946390151977539}
|
| 102 |
+
{"step": 102, "loss": 0.11110222823816197, "grad_norm": 0.15133707225322723, "lr": 7.815500501648653e-05, "epoch": 1, "t": 3.9376065731048584}
|
| 103 |
+
{"step": 103, "loss": 0.1274873402001611, "grad_norm": 0.1693481057882309, "lr": 7.769130192152538e-05, "epoch": 1, "t": 4.112709283828735}
|
| 104 |
+
{"step": 104, "loss": 0.08292039131361342, "grad_norm": 0.15406905114650726, "lr": 7.722413763244838e-05, "epoch": 1, "t": 3.7390711307525635}
|
| 105 |
+
{"step": 105, "loss": 0.11526951412018738, "grad_norm": 0.18622912466526031, "lr": 7.675357054110336e-05, "epoch": 1, "t": 3.727570056915283}
|
| 106 |
+
{"step": 106, "loss": 0.12006499917102575, "grad_norm": 0.17982099950313568, "lr": 7.627965946466166e-05, "epoch": 1, "t": 3.5842368602752686}
|
| 107 |
+
{"step": 107, "loss": 0.1186935144114395, "grad_norm": 0.14877408742904663, "lr": 7.580246363826621e-05, "epoch": 1, "t": 4.091334581375122}
|
| 108 |
+
{"step": 108, "loss": 0.13122232868610226, "grad_norm": 0.1457301527261734, "lr": 7.532204270762786e-05, "epoch": 1, "t": 3.881474256515503}
|
| 109 |
+
{"step": 109, "loss": 0.1334371601357872, "grad_norm": 0.14677388966083527, "lr": 7.483845672156998e-05, "epoch": 1, "t": 3.996276617050171}
|
| 110 |
+
{"step": 110, "loss": 0.13046903133467533, "grad_norm": 0.16596752405166626, "lr": 7.435176612452286e-05, "epoch": 1, "t": 4.019444227218628}
|
| 111 |
+
{"step": 111, "loss": 0.10941051667897489, "grad_norm": 0.17496438324451447, "lr": 7.386203174896872e-05, "epoch": 1, "t": 3.7645998001098633}
|
| 112 |
+
{"step": 112, "loss": 0.1244959343812367, "grad_norm": 0.16166579723358154, "lr": 7.336931480783801e-05, "epoch": 1, "t": 4.514106750488281}
|
| 113 |
+
{"step": 113, "loss": 0.11262510282407952, "grad_norm": 0.1646949052810669, "lr": 7.287367688685835e-05, "epoch": 1, "t": 4.528945446014404}
|
| 114 |
+
{"step": 114, "loss": 0.10489004908381282, "grad_norm": 0.18302619457244873, "lr": 7.237517993685678e-05, "epoch": 1, "t": 4.559117317199707}
|
| 115 |
+
{"step": 115, "loss": 0.10232535843065521, "grad_norm": 0.19442802667617798, "lr": 7.187388626601637e-05, "epoch": 1, "t": 4.564206600189209}
|
| 116 |
+
{"step": 116, "loss": 0.11103497537398845, "grad_norm": 0.16678892076015472, "lr": 7.136985853208824e-05, "epoch": 1, "t": 5.07295560836792}
|
| 117 |
+
{"step": 117, "loss": 0.1115818610558143, "grad_norm": 0.16402477025985718, "lr": 7.086315973455981e-05, "epoch": 1, "t": 5.139282703399658}
|
| 118 |
+
{"step": 118, "loss": 0.12158853422727228, "grad_norm": 0.1311653107404709, "lr": 7.035385320678036e-05, "epoch": 1, "t": 5.401008367538452}
|
| 119 |
+
{"step": 119, "loss": 0.09588072977468419, "grad_norm": 0.14086735248565674, "lr": 6.984200260804484e-05, "epoch": 1, "t": 4.952974557876587}
|
| 120 |
+
{"step": 120, "loss": 0.0912908763036104, "grad_norm": 0.16218799352645874, "lr": 6.932767191563703e-05, "epoch": 1, "t": 4.948731184005737}
|
| 121 |
+
{"step": 121, "loss": 0.13822233603085715, "grad_norm": 0.14316639304161072, "lr": 6.881092541683278e-05, "epoch": 1, "t": 5.22462797164917}
|
| 122 |
+
{"step": 122, "loss": 0.11881857142897552, "grad_norm": 0.1303873509168625, "lr": 6.829182770086474e-05, "epoch": 1, "t": 4.654219388961792}
|
| 123 |
+
{"step": 123, "loss": 0.12108342410574989, "grad_norm": 0.1579148918390274, "lr": 6.777044365084907e-05, "epoch": 1, "t": 3.9120402336120605}
|
| 124 |
+
{"step": 124, "loss": 0.07968569486767943, "grad_norm": 0.1335701048374176, "lr": 6.724683843567568e-05, "epoch": 1, "t": 3.8937900066375732}
|
| 125 |
+
{"step": 125, "loss": 0.11150920626381651, "grad_norm": 0.1818261593580246, "lr": 6.672107750186255e-05, "epoch": 1, "t": 3.5147409439086914}
|
| 126 |
+
{"step": 126, "loss": 0.14486833444633554, "grad_norm": 0.18314026296138763, "lr": 6.619322656537552e-05, "epoch": 1, "t": 3.928903579711914}
|
| 127 |
+
{"step": 127, "loss": 0.11722182175534124, "grad_norm": 0.13104470074176788, "lr": 6.566335160341424e-05, "epoch": 1, "t": 4.0404181480407715}
|
| 128 |
+
{"step": 128, "loss": 0.1222243271308053, "grad_norm": 0.12511242926120758, "lr": 6.513151884616556e-05, "epoch": 1, "t": 4.540865898132324}
|
| 129 |
+
{"step": 129, "loss": 0.1125568767840212, "grad_norm": 0.12010899931192398, "lr": 6.459779476852528e-05, "epoch": 1, "t": 4.350601434707642}
|
| 130 |
+
{"step": 130, "loss": 0.12215659848056286, "grad_norm": 0.13207314908504486, "lr": 6.406224608178932e-05, "epoch": 1, "t": 4.12429141998291}
|
| 131 |
+
{"step": 131, "loss": 0.11350722618544884, "grad_norm": 0.1452920138835907, "lr": 6.352493972531534e-05, "epoch": 1, "t": 4.59038233757019}
|
| 132 |
+
{"step": 132, "loss": 0.1238534170162562, "grad_norm": 0.13540174067020416, "lr": 6.298594285815584e-05, "epoch": 1, "t": 5.946590423583984}
|
| 133 |
+
{"step": 133, "loss": 0.10247356409141656, "grad_norm": 0.1491309404373169, "lr": 6.244532285066382e-05, "epoch": 1, "t": 5.886356592178345}
|
| 134 |
+
{"step": 134, "loss": 0.10514463958096298, "grad_norm": 0.16483132541179657, "lr": 6.190314727607196e-05, "epoch": 1, "t": 5.842749357223511}
|
| 135 |
+
{"step": 135, "loss": 0.11137263294971475, "grad_norm": 0.16692940890789032, "lr": 6.13594839020466e-05, "epoch": 1, "t": 5.612648963928223}
|
| 136 |
+
{"step": 136, "loss": 0.12202524957578044, "grad_norm": 0.1854080706834793, "lr": 6.0814400682217234e-05, "epoch": 1, "t": 5.886415243148804}
|
| 137 |
+
{"step": 137, "loss": 0.12197275685911982, "grad_norm": 0.16223259270191193, "lr": 6.026796574768288e-05, "epoch": 1, "t": 5.867288589477539}
|
| 138 |
+
{"step": 138, "loss": 0.0932202906921203, "grad_norm": 0.15473394095897675, "lr": 5.972024739849622e-05, "epoch": 1, "t": 5.83188009262085}
|
| 139 |
+
{"step": 139, "loss": 0.12311797054731247, "grad_norm": 0.13994963467121124, "lr": 5.917131409512663e-05, "epoch": 1, "t": 5.983316898345947}
|
| 140 |
+
{"step": 140, "loss": 0.13175687963900853, "grad_norm": 0.13232986629009247, "lr": 5.862123444990318e-05, "epoch": 1, "t": 5.916776895523071}
|
| 141 |
+
{"step": 141, "loss": 0.10179921986531704, "grad_norm": 0.14108580350875854, "lr": 5.807007721843861e-05, "epoch": 1, "t": 5.725611925125122}
|
| 142 |
+
{"step": 142, "loss": 0.12247693727065953, "grad_norm": 0.14080466330051422, "lr": 5.751791129103544e-05, "epoch": 1, "t": 5.3483970165252686}
|
| 143 |
+
{"step": 143, "loss": 0.13598552810749115, "grad_norm": 0.11519870162010193, "lr": 5.696480568407523e-05, "epoch": 1, "t": 6.234901666641235}
|
| 144 |
+
{"step": 144, "loss": 0.09938006266421522, "grad_norm": 0.14629501104354858, "lr": 5.6410829531392006e-05, "epoch": 1, "t": 5.59669828414917}
|
| 145 |
+
{"step": 145, "loss": 0.13709959031913002, "grad_norm": 0.1319306343793869, "lr": 5.585605207563124e-05, "epoch": 1, "t": 5.53915548324585}
|
| 146 |
+
{"step": 146, "loss": 0.11176312425211334, "grad_norm": 0.15270474553108215, "lr": 5.5300542659594854e-05, "epoch": 1, "t": 5.7169177532196045}
|
| 147 |
+
{"step": 147, "loss": 0.14579656180615397, "grad_norm": 0.1367747187614441, "lr": 5.47443707175741e-05, "epoch": 1, "t": 5.637811899185181}
|
| 148 |
+
{"step": 148, "loss": 0.12222280642755524, "grad_norm": 0.14530056715011597, "lr": 5.418760576667071e-05, "epoch": 1, "t": 5.36273717880249}
|
| 149 |
+
{"step": 149, "loss": 0.10807040342489989, "grad_norm": 0.14311166107654572, "lr": 5.3630317398107864e-05, "epoch": 1, "t": 5.1091697216033936}
|
| 150 |
+
{"step": 150, "loss": 0.10914952956936362, "grad_norm": 0.11722470074892044, "lr": 5.3072575268531835e-05, "epoch": 1, "t": 5.868841886520386}
|
| 151 |
+
{"step": 151, "loss": 0.1338075859704394, "grad_norm": 0.15515875816345215, "lr": 5.2514449091305375e-05, "epoch": 1, "t": 5.8782737255096436}
|
| 152 |
+
{"step": 152, "loss": 0.11667911126268332, "grad_norm": 0.13852252066135406, "lr": 5.195600862779421e-05, "epoch": 1, "t": 6.260565280914307}
|
| 153 |
+
{"step": 153, "loss": 0.12250399477943973, "grad_norm": 0.11874734610319138, "lr": 5.139732367864736e-05, "epoch": 1, "t": 5.6688232421875}
|
| 154 |
+
{"step": 154, "loss": 0.12234871340049323, "grad_norm": 0.14222052693367004, "lr": 5.083846407507263e-05, "epoch": 1, "t": 5.017601728439331}
|
| 155 |
+
{"step": 155, "loss": 0.1313176102610344, "grad_norm": 0.14096294343471527, "lr": 5.0279499670108245e-05, "epoch": 1, "t": 5.398028612136841}
|
| 156 |
+
{"step": 156, "loss": 0.10998567759641105, "grad_norm": 0.12560489773750305, "lr": 4.972050032989175e-05, "epoch": 1, "t": 5.881792306900024}
|
| 157 |
+
{"step": 157, "loss": 0.14253723142893687, "grad_norm": 0.12853525578975677, "lr": 4.9161535924927374e-05, "epoch": 1, "t": 6.006667375564575}
|
| 158 |
+
{"step": 158, "loss": 0.1362565209515793, "grad_norm": 0.18033531308174133, "lr": 4.860267632135265e-05, "epoch": 1, "t": 5.782114505767822}
|
| 159 |
+
{"step": 159, "loss": 0.12438243804136975, "grad_norm": 0.12671931087970734, "lr": 4.80439913722058e-05, "epoch": 1, "t": 6.095634698867798}
|
| 160 |
+
{"step": 160, "loss": 0.11695628953029918, "grad_norm": 0.13746081292629242, "lr": 4.748555090869464e-05, "epoch": 1, "t": 6.044499635696411}
|
| 161 |
+
{"step": 161, "loss": 0.10860840798888029, "grad_norm": 0.12572933733463287, "lr": 4.692742473146818e-05, "epoch": 1, "t": 6.383399248123169}
|
| 162 |
+
{"step": 162, "loss": 0.15660127597203813, "grad_norm": 0.16079489886760712, "lr": 4.636968260189214e-05, "epoch": 1, "t": 5.824679374694824}
|
| 163 |
+
{"step": 163, "loss": 0.10925060701854006, "grad_norm": 0.12640535831451416, "lr": 4.5812394233329305e-05, "epoch": 1, "t": 5.920647144317627}
|
| 164 |
+
{"step": 164, "loss": 0.13847018529098065, "grad_norm": 0.13318540155887604, "lr": 4.525562928242592e-05, "epoch": 1, "t": 6.111425399780273}
|
| 165 |
+
{"step": 165, "loss": 0.11645202288874017, "grad_norm": 0.14153780043125153, "lr": 4.4699457340405164e-05, "epoch": 1, "t": 5.860616683959961}
|
| 166 |
+
{"step": 166, "loss": 0.10572560891730859, "grad_norm": 0.14335797727108002, "lr": 4.414394792436877e-05, "epoch": 1, "t": 5.821516036987305}
|
| 167 |
+
{"step": 167, "loss": 0.12092660270690916, "grad_norm": 0.13772067427635193, "lr": 4.3589170468607985e-05, "epoch": 1, "t": 5.985003709793091}
|
| 168 |
+
{"step": 168, "loss": 0.11941248052510059, "grad_norm": 0.1498883068561554, "lr": 4.3035194315924785e-05, "epoch": 1, "t": 5.9461143016815186}
|
| 169 |
+
{"step": 169, "loss": 0.09770603653573595, "grad_norm": 0.1489756852388382, "lr": 4.248208870896456e-05, "epoch": 1, "t": 5.789595365524292}
|
| 170 |
+
{"step": 170, "loss": 0.12906971163680264, "grad_norm": 0.12248232960700989, "lr": 4.192992278156141e-05, "epoch": 1, "t": 5.774706125259399}
|
| 171 |
+
{"step": 171, "loss": 0.10495811824118934, "grad_norm": 0.12584099173545837, "lr": 4.1378765550096835e-05, "epoch": 1, "t": 4.037522315979004}
|
| 172 |
+
{"step": 172, "loss": 0.15494878209609153, "grad_norm": 0.146627277135849, "lr": 4.082868590487339e-05, "epoch": 1, "t": 4.370777130126953}
|
| 173 |
+
{"step": 173, "loss": 0.13606241378445846, "grad_norm": 0.13438133895397186, "lr": 4.027975260150381e-05, "epoch": 1, "t": 4.2283103466033936}
|
| 174 |
+
{"step": 174, "loss": 0.11738633436514131, "grad_norm": 0.1292438954114914, "lr": 3.973203425231715e-05, "epoch": 1, "t": 4.1627538204193115}
|
| 175 |
+
{"step": 175, "loss": 0.09793814501262667, "grad_norm": 0.14427684247493744, "lr": 3.918559931778277e-05, "epoch": 1, "t": 3.76477313041687}
|
| 176 |
+
{"step": 176, "loss": 0.1134855055438422, "grad_norm": 0.13561420142650604, "lr": 3.8640516097953405e-05, "epoch": 1, "t": 4.130195379257202}
|
| 177 |
+
{"step": 177, "loss": 0.10737729666508591, "grad_norm": 0.13862013816833496, "lr": 3.809685272392804e-05, "epoch": 1, "t": 4.025635480880737}
|
| 178 |
+
{"step": 178, "loss": 0.14615545516486445, "grad_norm": 0.15243802964687347, "lr": 3.755467714933619e-05, "epoch": 1, "t": 4.069834470748901}
|
| 179 |
+
{"step": 179, "loss": 0.13059986952499916, "grad_norm": 0.15739114582538605, "lr": 3.701405714184416e-05, "epoch": 1, "t": 4.0566606521606445}
|
| 180 |
+
{"step": 180, "loss": 0.1250554776811917, "grad_norm": 0.1449676752090454, "lr": 3.647506027468467e-05, "epoch": 1, "t": 3.888803482055664}
|
| 181 |
+
{"step": 181, "loss": 0.1339609112250817, "grad_norm": 0.11792629957199097, "lr": 3.59377539182107e-05, "epoch": 1, "t": 5.0446813106536865}
|
| 182 |
+
{"step": 182, "loss": 0.0964094779671148, "grad_norm": 0.16920435428619385, "lr": 3.5402205231474736e-05, "epoch": 1, "t": 3.5308339595794678}
|
| 183 |
+
{"step": 183, "loss": 0.0841054949191732, "grad_norm": 0.158295139670372, "lr": 3.486848115383445e-05, "epoch": 1, "t": 3.699392557144165}
|
| 184 |
+
{"step": 184, "loss": 0.12007146771255114, "grad_norm": 0.1564946174621582, "lr": 3.4336648396585776e-05, "epoch": 1, "t": 3.737936496734619}
|
| 185 |
+
{"step": 185, "loss": 0.07932232825931569, "grad_norm": 0.12786969542503357, "lr": 3.380677343462447e-05, "epoch": 1, "t": 3.772343635559082}
|
| 186 |
+
{"step": 186, "loss": 0.11378491893440214, "grad_norm": 0.16465850174427032, "lr": 3.327892249813745e-05, "epoch": 1, "t": 3.755070447921753}
|
| 187 |
+
{"step": 187, "loss": 0.0909288471611379, "grad_norm": 0.13516880571842194, "lr": 3.275316156432434e-05, "epoch": 1, "t": 3.848257541656494}
|
| 188 |
+
{"step": 188, "loss": 0.14310075986846685, "grad_norm": 0.1411554217338562, "lr": 3.2229556349150945e-05, "epoch": 1, "t": 4.250707626342773}
|
| 189 |
+
{"step": 189, "loss": 0.11388757989347517, "grad_norm": 0.12976497411727905, "lr": 3.170817229913526e-05, "epoch": 1, "t": 4.144380807876587}
|
| 190 |
+
{"step": 190, "loss": 0.1265786284890618, "grad_norm": 0.1473735123872757, "lr": 3.118907458316722e-05, "epoch": 1, "t": 4.082712888717651}
|
| 191 |
+
{"step": 191, "loss": 0.13427961269346605, "grad_norm": 0.15153977274894714, "lr": 3.067232808436299e-05, "epoch": 1, "t": 4.0815465450286865}
|
| 192 |
+
{"step": 192, "loss": 0.10992266317146325, "grad_norm": 0.15019188821315765, "lr": 3.0157997391955172e-05, "epoch": 1, "t": 4.350877046585083}
|
| 193 |
+
{"step": 193, "loss": 0.1572723716825705, "grad_norm": 0.12622733414173126, "lr": 2.964614679321966e-05, "epoch": 1, "t": 4.8909242153167725}
|
| 194 |
+
{"step": 194, "loss": 0.11537074267244213, "grad_norm": 0.15508973598480225, "lr": 2.913684026544021e-05, "epoch": 1, "t": 4.080131530761719}
|
| 195 |
+
{"step": 195, "loss": 0.12344161256182586, "grad_norm": 0.16149987280368805, "lr": 2.8630141467911775e-05, "epoch": 1, "t": 3.9379467964172363}
|
| 196 |
+
{"step": 196, "loss": 0.13281372677716102, "grad_norm": 0.15665747225284576, "lr": 2.812611373398365e-05, "epoch": 1, "t": 3.7952842712402344}
|
| 197 |
+
{"step": 197, "loss": 0.16175826833642656, "grad_norm": 0.1399657279253006, "lr": 2.762482006314324e-05, "epoch": 1, "t": 4.240842580795288}
|
| 198 |
+
{"step": 198, "loss": 0.12268284408389486, "grad_norm": 0.15245409309864044, "lr": 2.712632311314165e-05, "epoch": 1, "t": 4.2339842319488525}
|
| 199 |
+
{"step": 199, "loss": 0.10661711567550075, "grad_norm": 0.1496862769126892, "lr": 2.6630685192161992e-05, "epoch": 1, "t": 3.873534679412842}
|
| 200 |
+
{"step": 200, "loss": 0.10718047681134313, "grad_norm": 0.16035036742687225, "lr": 2.6137968251031287e-05, "epoch": 1, "t": 3.512685775756836}
|
| 201 |
+
{"step": 201, "loss": 0.09733861004859186, "grad_norm": 0.1342783272266388, "lr": 2.5648233875477157e-05, "epoch": 1, "t": 3.726299285888672}
|
| 202 |
+
{"step": 202, "loss": 0.10710803163853956, "grad_norm": 0.14960817992687225, "lr": 2.5161543278430054e-05, "epoch": 1, "t": 3.5803110599517822}
|
| 203 |
+
{"step": 203, "loss": 0.12856140146460426, "grad_norm": 0.14114242792129517, "lr": 2.4677957292372167e-05, "epoch": 1, "t": 3.8162992000579834}
|
| 204 |
+
{"step": 204, "loss": 0.09950678060929327, "grad_norm": 0.13951867818832397, "lr": 2.419753636173379e-05, "epoch": 1, "t": 3.4555318355560303}
|
| 205 |
+
{"step": 205, "loss": 0.1192101331395427, "grad_norm": 0.15045635402202606, "lr": 2.3720340535338348e-05, "epoch": 1, "t": 3.754653215408325}
|
| 206 |
+
{"step": 206, "loss": 0.10550485139810536, "grad_norm": 0.1489887237548828, "lr": 2.3246429458896634e-05, "epoch": 1, "t": 3.4955015182495117}
|
| 207 |
+
{"step": 207, "loss": 0.1457673885732749, "grad_norm": 0.1404895782470703, "lr": 2.2775862367551644e-05, "epoch": 1, "t": 4.137854814529419}
|
| 208 |
+
{"step": 208, "loss": 0.09591870073622033, "grad_norm": 0.12496516108512878, "lr": 2.2308698078474645e-05, "epoch": 1, "t": 3.657066583633423}
|
| 209 |
+
{"step": 209, "loss": 0.11847254509543413, "grad_norm": 0.1287345588207245, "lr": 2.1844994983513467e-05, "epoch": 1, "t": 3.7544782161712646}
|
| 210 |
+
{"step": 210, "loss": 0.0979276126730907, "grad_norm": 0.14218559861183167, "lr": 2.1384811041894055e-05, "epoch": 1, "t": 3.231539726257324}
|
| 211 |
+
{"step": 211, "loss": 0.09428850225518377, "grad_norm": 0.1390930414199829, "lr": 2.0928203772975917e-05, "epoch": 1, "t": 3.346951484680176}
|
| 212 |
+
{"step": 212, "loss": 0.12567077994677361, "grad_norm": 0.17330411076545715, "lr": 2.0475230249062725e-05, "epoch": 1, "t": 3.1886229515075684}
|
| 213 |
+
{"step": 213, "loss": 0.09782922963854526, "grad_norm": 0.13580435514450073, "lr": 2.0025947088268717e-05, "epoch": 1, "t": 3.3848838806152344}
|
| 214 |
+
{"step": 214, "loss": 0.13602645966365862, "grad_norm": 0.17799289524555206, "lr": 1.958041044744186e-05, "epoch": 1, "t": 3.171832799911499}
|
| 215 |
+
{"step": 215, "loss": 0.10618019406428078, "grad_norm": 0.1461976170539856, "lr": 1.9138676015144764e-05, "epoch": 1, "t": 3.309781551361084}
|
| 216 |
+
{"step": 216, "loss": 0.09441785260250693, "grad_norm": 0.13548174500465393, "lr": 1.870079900469392e-05, "epoch": 1, "t": 3.8321173191070557}
|
| 217 |
+
{"step": 217, "loss": 0.09281340758928955, "grad_norm": 0.14920149743556976, "lr": 1.8266834147258578e-05, "epoch": 1, "t": 3.2834720611572266}
|
| 218 |
+
{"step": 218, "loss": 0.13122085211118584, "grad_norm": 0.17590315639972687, "lr": 1.7836835685019733e-05, "epoch": 1, "t": 3.6829259395599365}
|
| 219 |
+
{"step": 219, "loss": 0.14349004887697991, "grad_norm": 0.13497987389564514, "lr": 1.741085736439031e-05, "epoch": 1, "t": 4.1920435428619385}
|
| 220 |
+
{"step": 220, "loss": 0.12411245391902115, "grad_norm": 0.1270938217639923, "lr": 1.698895242929725e-05, "epoch": 1, "t": 4.421627044677734}
|
| 221 |
+
{"step": 221, "loss": 0.10102598842648428, "grad_norm": 0.15080025792121887, "lr": 1.6571173614526507e-05, "epoch": 1, "t": 3.919564962387085}
|
| 222 |
+
{"step": 222, "loss": 0.1344824544603108, "grad_norm": 0.13348388671875, "lr": 1.6157573139131527e-05, "epoch": 1, "t": 4.353817939758301}
|
| 223 |
+
{"step": 223, "loss": 0.11996158967234363, "grad_norm": 0.1332148164510727, "lr": 1.5748202699906335e-05, "epoch": 1, "t": 4.168254852294922}
|
| 224 |
+
{"step": 224, "loss": 0.12410005547497649, "grad_norm": 0.12707005441188812, "lr": 1.534311346492381e-05, "epoch": 1, "t": 5.450979232788086}
|
| 225 |
+
{"step": 225, "loss": 0.10171063207279571, "grad_norm": 0.17476694285869598, "lr": 1.4942356067140162e-05, "epoch": 1, "t": 4.6596598625183105}
|
| 226 |
+
{"step": 226, "loss": 0.08603258519586866, "grad_norm": 0.10637104511260986, "lr": 1.454598059806609e-05, "epoch": 1, "t": 6.157179594039917}
|
| 227 |
+
{"step": 227, "loss": 0.14213648385105823, "grad_norm": 0.1354423314332962, "lr": 1.4154036601505832e-05, "epoch": 1, "t": 5.06745457649231}
|
| 228 |
+
{"step": 228, "loss": 0.12114619999198262, "grad_norm": 0.12042339146137238, "lr": 1.376657306736453e-05, "epoch": 1, "t": 5.55302882194519}
|
| 229 |
+
{"step": 229, "loss": 0.14443165937833533, "grad_norm": 0.11341030895709991, "lr": 1.3383638425524908e-05, "epoch": 1, "t": 6.407577276229858}
|
| 230 |
+
{"step": 230, "loss": 0.12206946873519237, "grad_norm": 0.13370126485824585, "lr": 1.3005280539793907e-05, "epoch": 1, "t": 5.899457931518555}
|
| 231 |
+
{"step": 231, "loss": 0.14863756474731973, "grad_norm": 0.14756746590137482, "lr": 1.2631546701920071e-05, "epoch": 1, "t": 5.333111047744751}
|
| 232 |
+
{"step": 232, "loss": 0.13004562389244703, "grad_norm": 0.1426292061805725, "lr": 1.2262483625682513e-05, "epoch": 1, "t": 5.817943572998047}
|
| 233 |
+
{"step": 233, "loss": 0.11405566817683196, "grad_norm": 0.1313171237707138, "lr": 1.1898137441051982e-05, "epoch": 1, "t": 5.368566036224365}
|
| 234 |
+
{"step": 234, "loss": 0.09286493925544305, "grad_norm": 0.12980012595653534, "lr": 1.1538553688425003e-05, "epoch": 1, "t": 6.268900394439697}
|
| 235 |
+
{"step": 235, "loss": 0.09465315181374723, "grad_norm": 0.1298038810491562, "lr": 1.1183777312931748e-05, "epoch": 1, "t": 5.635498762130737}
|
| 236 |
+
{"step": 236, "loss": 0.10800215024922183, "grad_norm": 0.14482581615447998, "lr": 1.0833852658818166e-05, "epoch": 1, "t": 4.18051290512085}
|
| 237 |
+
{"step": 237, "loss": 0.08360361414277363, "grad_norm": 0.14186900854110718, "lr": 1.0488823463903342e-05, "epoch": 1, "t": 3.251851797103882}
|
| 238 |
+
{"step": 238, "loss": 0.1254756912914426, "grad_norm": 0.1392965465784073, "lr": 1.0148732854112619e-05, "epoch": 1, "t": 3.7237584590911865}
|
| 239 |
+
{"step": 239, "loss": 0.14220688001573326, "grad_norm": 0.1264325976371765, "lr": 9.81362333808718e-06, "epoch": 1, "t": 4.819896221160889}
|
| 240 |
+
{"step": 240, "loss": 0.12136467044813183, "grad_norm": 0.1665497124195099, "lr": 9.483536801870834e-06, "epoch": 1, "t": 3.3412585258483887}
|
| 241 |
+
{"step": 241, "loss": 0.12490963203455108, "grad_norm": 0.15111815929412842, "lr": 9.158514503674543e-06, "epoch": 1, "t": 3.4438259601593018}
|
| 242 |
+
{"step": 242, "loss": 0.12017831168417333, "grad_norm": 0.1290101557970047, "lr": 8.838597068719518e-06, "epoch": 1, "t": 4.016022205352783}
|
| 243 |
+
{"step": 243, "loss": 0.09489677106351789, "grad_norm": 0.11944586038589478, "lr": 8.523824484159349e-06, "epoch": 1, "t": 3.92342472076416}
|
| 244 |
+
{"step": 244, "loss": 0.1243680059571503, "grad_norm": 0.1327151656150818, "lr": 8.21423609408199e-06, "epoch": 1, "t": 3.734070062637329}
|
| 245 |
+
{"step": 245, "loss": 0.11684684307067719, "grad_norm": 0.15995414555072784, "lr": 7.90987059459195e-06, "epoch": 1, "t": 3.4899020195007324}
|
| 246 |
+
{"step": 246, "loss": 0.12360585496278272, "grad_norm": 0.1631646752357483, "lr": 7.610766028973709e-06, "epoch": 1, "t": 4.229679107666016}
|
| 247 |
+
{"step": 247, "loss": 0.13372521546217367, "grad_norm": 0.15672412514686584, "lr": 7.3169597829365165e-06, "epoch": 1, "t": 4.729381084442139}
|
| 248 |
+
{"step": 248, "loss": 0.11260705068762944, "grad_norm": 0.13472774624824524, "lr": 7.028488579941506e-06, "epoch": 1, "t": 4.950815916061401}
|
| 249 |
+
{"step": 249, "loss": 0.123039937727079, "grad_norm": 0.11839025467634201, "lr": 6.745388476611553e-06, "epoch": 1, "t": 5.320116281509399}
|
| 250 |
+
{"step": 250, "loss": 0.11276406303449972, "grad_norm": 0.1768644005060196, "lr": 6.467694858224488e-06, "epoch": 1, "t": 4.611212730407715}
|
| 251 |
+
{"step": 251, "loss": 0.14011463678387218, "grad_norm": 0.13780367374420166, "lr": 6.1954424342902e-06, "epoch": 1, "t": 4.852539777755737}
|
| 252 |
+
{"step": 252, "loss": 0.09068210728985011, "grad_norm": 0.13530129194259644, "lr": 5.928665234212233e-06, "epoch": 1, "t": 4.842432498931885}
|
| 253 |
+
{"step": 253, "loss": 0.10120193497090002, "grad_norm": 0.11738213896751404, "lr": 5.66739660303437e-06, "epoch": 1, "t": 5.2902562618255615}
|
| 254 |
+
{"step": 254, "loss": 0.09851244988383807, "grad_norm": 0.1303517073392868, "lr": 5.411669197272795e-06, "epoch": 1, "t": 5.693821430206299}
|
| 255 |
+
{"step": 255, "loss": 0.10302092982955194, "grad_norm": 0.1410181075334549, "lr": 5.161514980834231e-06, "epoch": 1, "t": 5.209938049316406}
|
| 256 |
+
{"step": 256, "loss": 0.13269496118584523, "grad_norm": 0.15145458281040192, "lr": 4.916965221020753e-06, "epoch": 1, "t": 5.3764777183532715}
|
| 257 |
+
{"step": 257, "loss": 0.10233825561732993, "grad_norm": 0.13183993101119995, "lr": 4.678050484621615e-06, "epoch": 1, "t": 5.46918249130249}
|
| 258 |
+
{"step": 258, "loss": 0.13216298849471492, "grad_norm": 0.13832338154315948, "lr": 4.444800634092616e-06, "epoch": 1, "t": 4.821920871734619}
|
| 259 |
+
{"step": 259, "loss": 0.12163336211593835, "grad_norm": 0.15184752643108368, "lr": 4.217244823823546e-06, "epoch": 1, "t": 3.7082533836364746}
|
| 260 |
+
{"step": 260, "loss": 0.15002393293209007, "grad_norm": 0.15476936101913452, "lr": 3.995411496494134e-06, "epoch": 1, "t": 3.559565782546997}
|
| 261 |
+
{"step": 261, "loss": 0.09959995404541193, "grad_norm": 0.12526850402355194, "lr": 3.7793283795188984e-06, "epoch": 1, "t": 3.98399019241333}
|
| 262 |
+
{"step": 262, "loss": 0.11795829714304533, "grad_norm": 0.16237178444862366, "lr": 3.56902248158148e-06, "epoch": 1, "t": 3.2178425788879395}
|
| 263 |
+
{"step": 263, "loss": 0.14876536610499433, "grad_norm": 0.15125760436058044, "lr": 3.364520089258727e-06, "epoch": 1, "t": 3.726151704788208}
|
| 264 |
+
{"step": 264, "loss": 0.1307659820311301, "grad_norm": 0.13269969820976257, "lr": 3.165846763735153e-06, "epoch": 1, "t": 4.523290634155273}
|
| 265 |
+
{"step": 265, "loss": 0.09039410618621412, "grad_norm": 0.15942604839801788, "lr": 2.973027337607892e-06, "epoch": 1, "t": 3.1473255157470703}
|
| 266 |
+
{"step": 266, "loss": 0.1279116300673159, "grad_norm": 0.16423822939395905, "lr": 2.7860859117828987e-06, "epoch": 1, "t": 3.282160520553589}
|
| 267 |
+
{"step": 267, "loss": 0.08462305770535225, "grad_norm": 0.14013220369815826, "lr": 2.605045852462473e-06, "epoch": 1, "t": 3.2357003688812256}
|
| 268 |
+
{"step": 268, "loss": 0.09663602861307435, "grad_norm": 0.17577238380908966, "lr": 2.429929788224722e-06, "epoch": 1, "t": 3.242727756500244}
|
| 269 |
+
{"step": 269, "loss": 0.14687636517096836, "grad_norm": 0.16681092977523804, "lr": 2.2607596071951286e-06, "epoch": 1, "t": 4.1681108474731445}
|
| 270 |
+
{"step": 270, "loss": 0.11895260757699186, "grad_norm": 0.12868773937225342, "lr": 2.097556454310701e-06, "epoch": 1, "t": 4.703555583953857}
|
| 271 |
+
{"step": 271, "loss": 0.10439320327601696, "grad_norm": 0.1587361991405487, "lr": 1.940340728677059e-06, "epoch": 1, "t": 4.763911247253418}
|
| 272 |
+
{"step": 272, "loss": 0.11546959674368708, "grad_norm": 0.15578536689281464, "lr": 1.789132081018674e-06, "epoch": 1, "t": 4.900822877883911}
|
| 273 |
+
{"step": 273, "loss": 0.12891419631590156, "grad_norm": 0.15120558440685272, "lr": 1.6439494112227172e-06, "epoch": 1, "t": 4.793020725250244}
|
| 274 |
+
{"step": 274, "loss": 0.15081281376979955, "grad_norm": 0.12745462357997894, "lr": 1.5048108659766691e-06, "epoch": 1, "t": 5.244046688079834}
|
| 275 |
+
{"step": 275, "loss": 0.10252382758882515, "grad_norm": 0.12532322108745575, "lr": 1.3717338365001942e-06, "epoch": 1, "t": 5.692639589309692}
|
| 276 |
+
{"step": 276, "loss": 0.1052388965172062, "grad_norm": 0.17532673478126526, "lr": 1.2447349563713184e-06, "epoch": 1, "t": 4.433902025222778}
|
| 277 |
+
{"step": 277, "loss": 0.12041700529015582, "grad_norm": 0.14770610630512238, "lr": 1.1238300994473983e-06, "epoch": 1, "t": 4.664698839187622}
|
| 278 |
+
{"step": 278, "loss": 0.0990669545895036, "grad_norm": 0.13431894779205322, "lr": 1.0090343778809908e-06, "epoch": 1, "t": 4.974935054779053}
|
| 279 |
+
{"step": 279, "loss": 0.10497822829783265, "grad_norm": 0.12812137603759766, "lr": 9.003621402309814e-07, "epoch": 1, "t": 4.980578899383545}
|
| 280 |
+
{"step": 280, "loss": 0.08761405289810818, "grad_norm": 0.1214241236448288, "lr": 7.97826969669102e-07, "epoch": 1, "t": 5.631446123123169}
|
| 281 |
+
{"step": 281, "loss": 0.12079584596448073, "grad_norm": 0.13822419941425323, "lr": 7.014416822821556e-07, "epoch": 1, "t": 5.562159299850464}
|
| 282 |
+
{"step": 282, "loss": 0.11831246908856181, "grad_norm": 0.15493813157081604, "lr": 6.112183254700865e-07, "epoch": 1, "t": 5.366363763809204}
|
| 283 |
+
{"step": 283, "loss": 0.0857640213671429, "grad_norm": 0.1580144762992859, "lr": 5.271681764401848e-07, "epoch": 1, "t": 5.280276775360107}
|
| 284 |
+
{"step": 284, "loss": 0.1355166469985603, "grad_norm": 0.14353851974010468, "lr": 4.493017407975086e-07, "epoch": 1, "t": 5.418309450149536}
|
| 285 |
+
{"step": 285, "loss": 0.1259592493205691, "grad_norm": 0.15438403189182281, "lr": 3.7762875123173445e-07, "epoch": 1, "t": 5.79780912399292}
|
| 286 |
+
{"step": 286, "loss": 0.11926719573508451, "grad_norm": 0.12321104109287262, "lr": 3.1215816630071335e-07, "epoch": 1, "t": 5.482792854309082}
|
| 287 |
+
{"step": 287, "loss": 0.11562612069711753, "grad_norm": 0.14055198431015015, "lr": 2.528981693106558e-07, "epoch": 1, "t": 5.365353107452393}
|
| 288 |
+
{"step": 288, "loss": 0.12881073325033107, "grad_norm": 0.13906581699848175, "lr": 1.9985616729332747e-07, "epoch": 1, "t": 5.620147943496704}
|
| 289 |
+
{"step": 289, "loss": 0.13206261805223798, "grad_norm": 0.14486300945281982, "lr": 1.530387900802177e-07, "epoch": 1, "t": 4.856403112411499}
|
| 290 |
+
{"step": 290, "loss": 0.10525038197031221, "grad_norm": 0.1403212547302246, "lr": 1.1245188947384134e-07, "epoch": 1, "t": 4.65713906288147}
|
| 291 |
+
{"step": 291, "loss": 0.14182153682170592, "grad_norm": 0.1460697203874588, "lr": 7.81005385163458e-08, "epoch": 1, "t": 4.921620845794678}
|
| 292 |
+
{"step": 292, "loss": 0.1270982598292454, "grad_norm": 0.13338321447372437, "lr": 4.998903085539075e-08, "epoch": 1, "t": 5.010462284088135}
|
| 293 |
+
{"step": 293, "loss": 0.11927410909358507, "grad_norm": 0.1483747512102127, "lr": 2.8120880207493928e-08, "epoch": 1, "t": 4.673955202102661}
|
| 294 |
+
{"step": 294, "loss": 0.08697800355922135, "grad_norm": 0.15373092889785767, "lr": 1.2498819918843607e-08, "epoch": 1, "t": 4.913345813751221}
|
| 295 |
+
{"step": 295, "loss": 0.0985233740531525, "grad_norm": 0.13708575069904327, "lr": 3.1248026236274652e-09, "epoch": 1, "t": 5.145484924316406}
|
| 296 |
+
{"step": 296, "loss": 0.10419206268442525, "grad_norm": 0.15778356790542603, "lr": 0.0, "epoch": 1, "t": 5.274567604064941}
|
qwen3-14b/rand-s0/README.md
ADDED
|
@@ -0,0 +1,206 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: /workspace-vast/jfonsecarivera/rlw-hf/hub/models--Qwen--Qwen3-14B/snapshots/40c069824f4251a91eefaf281ebe4c544efd3e18
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- base_model:adapter:/workspace-vast/jfonsecarivera/rlw-hf/hub/models--Qwen--Qwen3-14B/snapshots/40c069824f4251a91eefaf281ebe4c544efd3e18
|
| 6 |
+
- lora
|
| 7 |
+
- transformers
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# Model Card for Model ID
|
| 11 |
+
|
| 12 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
## Model Details
|
| 17 |
+
|
| 18 |
+
### Model Description
|
| 19 |
+
|
| 20 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
- **Developed by:** [More Information Needed]
|
| 25 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 26 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 27 |
+
- **Model type:** [More Information Needed]
|
| 28 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 29 |
+
- **License:** [More Information Needed]
|
| 30 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 31 |
+
|
| 32 |
+
### Model Sources [optional]
|
| 33 |
+
|
| 34 |
+
<!-- Provide the basic links for the model. -->
|
| 35 |
+
|
| 36 |
+
- **Repository:** [More Information Needed]
|
| 37 |
+
- **Paper [optional]:** [More Information Needed]
|
| 38 |
+
- **Demo [optional]:** [More Information Needed]
|
| 39 |
+
|
| 40 |
+
## Uses
|
| 41 |
+
|
| 42 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 43 |
+
|
| 44 |
+
### Direct Use
|
| 45 |
+
|
| 46 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 47 |
+
|
| 48 |
+
[More Information Needed]
|
| 49 |
+
|
| 50 |
+
### Downstream Use [optional]
|
| 51 |
+
|
| 52 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 53 |
+
|
| 54 |
+
[More Information Needed]
|
| 55 |
+
|
| 56 |
+
### Out-of-Scope Use
|
| 57 |
+
|
| 58 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 59 |
+
|
| 60 |
+
[More Information Needed]
|
| 61 |
+
|
| 62 |
+
## Bias, Risks, and Limitations
|
| 63 |
+
|
| 64 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 65 |
+
|
| 66 |
+
[More Information Needed]
|
| 67 |
+
|
| 68 |
+
### Recommendations
|
| 69 |
+
|
| 70 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 71 |
+
|
| 72 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 73 |
+
|
| 74 |
+
## How to Get Started with the Model
|
| 75 |
+
|
| 76 |
+
Use the code below to get started with the model.
|
| 77 |
+
|
| 78 |
+
[More Information Needed]
|
| 79 |
+
|
| 80 |
+
## Training Details
|
| 81 |
+
|
| 82 |
+
### Training Data
|
| 83 |
+
|
| 84 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 85 |
+
|
| 86 |
+
[More Information Needed]
|
| 87 |
+
|
| 88 |
+
### Training Procedure
|
| 89 |
+
|
| 90 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 91 |
+
|
| 92 |
+
#### Preprocessing [optional]
|
| 93 |
+
|
| 94 |
+
[More Information Needed]
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
#### Training Hyperparameters
|
| 98 |
+
|
| 99 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 100 |
+
|
| 101 |
+
#### Speeds, Sizes, Times [optional]
|
| 102 |
+
|
| 103 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 104 |
+
|
| 105 |
+
[More Information Needed]
|
| 106 |
+
|
| 107 |
+
## Evaluation
|
| 108 |
+
|
| 109 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 110 |
+
|
| 111 |
+
### Testing Data, Factors & Metrics
|
| 112 |
+
|
| 113 |
+
#### Testing Data
|
| 114 |
+
|
| 115 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 116 |
+
|
| 117 |
+
[More Information Needed]
|
| 118 |
+
|
| 119 |
+
#### Factors
|
| 120 |
+
|
| 121 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 122 |
+
|
| 123 |
+
[More Information Needed]
|
| 124 |
+
|
| 125 |
+
#### Metrics
|
| 126 |
+
|
| 127 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 128 |
+
|
| 129 |
+
[More Information Needed]
|
| 130 |
+
|
| 131 |
+
### Results
|
| 132 |
+
|
| 133 |
+
[More Information Needed]
|
| 134 |
+
|
| 135 |
+
#### Summary
|
| 136 |
+
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
## Model Examination [optional]
|
| 140 |
+
|
| 141 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 142 |
+
|
| 143 |
+
[More Information Needed]
|
| 144 |
+
|
| 145 |
+
## Environmental Impact
|
| 146 |
+
|
| 147 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 148 |
+
|
| 149 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 150 |
+
|
| 151 |
+
- **Hardware Type:** [More Information Needed]
|
| 152 |
+
- **Hours used:** [More Information Needed]
|
| 153 |
+
- **Cloud Provider:** [More Information Needed]
|
| 154 |
+
- **Compute Region:** [More Information Needed]
|
| 155 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 156 |
+
|
| 157 |
+
## Technical Specifications [optional]
|
| 158 |
+
|
| 159 |
+
### Model Architecture and Objective
|
| 160 |
+
|
| 161 |
+
[More Information Needed]
|
| 162 |
+
|
| 163 |
+
### Compute Infrastructure
|
| 164 |
+
|
| 165 |
+
[More Information Needed]
|
| 166 |
+
|
| 167 |
+
#### Hardware
|
| 168 |
+
|
| 169 |
+
[More Information Needed]
|
| 170 |
+
|
| 171 |
+
#### Software
|
| 172 |
+
|
| 173 |
+
[More Information Needed]
|
| 174 |
+
|
| 175 |
+
## Citation [optional]
|
| 176 |
+
|
| 177 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 178 |
+
|
| 179 |
+
**BibTeX:**
|
| 180 |
+
|
| 181 |
+
[More Information Needed]
|
| 182 |
+
|
| 183 |
+
**APA:**
|
| 184 |
+
|
| 185 |
+
[More Information Needed]
|
| 186 |
+
|
| 187 |
+
## Glossary [optional]
|
| 188 |
+
|
| 189 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 190 |
+
|
| 191 |
+
[More Information Needed]
|
| 192 |
+
|
| 193 |
+
## More Information [optional]
|
| 194 |
+
|
| 195 |
+
[More Information Needed]
|
| 196 |
+
|
| 197 |
+
## Model Card Authors [optional]
|
| 198 |
+
|
| 199 |
+
[More Information Needed]
|
| 200 |
+
|
| 201 |
+
## Model Card Contact
|
| 202 |
+
|
| 203 |
+
[More Information Needed]
|
| 204 |
+
### Framework versions
|
| 205 |
+
|
| 206 |
+
- PEFT 0.17.1
|
qwen3-14b/rand-s0/adapter_config.json
ADDED
|
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": {
|
| 4 |
+
"base_model_class": "Qwen3ForCausalLM",
|
| 5 |
+
"parent_library": "transformers.models.qwen3.modeling_qwen3"
|
| 6 |
+
},
|
| 7 |
+
"base_model_name_or_path": "/workspace-vast/jfonsecarivera/rlw-hf/hub/models--Qwen--Qwen3-14B/snapshots/40c069824f4251a91eefaf281ebe4c544efd3e18",
|
| 8 |
+
"bias": "none",
|
| 9 |
+
"corda_config": null,
|
| 10 |
+
"eva_config": null,
|
| 11 |
+
"exclude_modules": null,
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": true,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layer_replication": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"layers_to_transform": null,
|
| 18 |
+
"loftq_config": {},
|
| 19 |
+
"lora_alpha": 128,
|
| 20 |
+
"lora_bias": false,
|
| 21 |
+
"lora_dropout": 0.05,
|
| 22 |
+
"megatron_config": null,
|
| 23 |
+
"megatron_core": "megatron.core",
|
| 24 |
+
"modules_to_save": null,
|
| 25 |
+
"peft_type": "LORA",
|
| 26 |
+
"qalora_group_size": 16,
|
| 27 |
+
"r": 64,
|
| 28 |
+
"rank_pattern": {},
|
| 29 |
+
"revision": null,
|
| 30 |
+
"target_modules": [
|
| 31 |
+
"k_proj",
|
| 32 |
+
"up_proj",
|
| 33 |
+
"down_proj",
|
| 34 |
+
"o_proj",
|
| 35 |
+
"v_proj",
|
| 36 |
+
"q_proj",
|
| 37 |
+
"gate_proj"
|
| 38 |
+
],
|
| 39 |
+
"target_parameters": null,
|
| 40 |
+
"task_type": null,
|
| 41 |
+
"trainable_token_indices": null,
|
| 42 |
+
"use_dora": false,
|
| 43 |
+
"use_qalora": false,
|
| 44 |
+
"use_rslora": false
|
| 45 |
+
}
|
qwen3-14b/rand-s0/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:13a6a95d889319ff0e4aa44980555556a96f257404587b2d8728084d5bef0cca
|
| 3 |
+
size 1027680256
|
qwen3-14b/rand-s0/config.json
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "/workspace-vast/jfonsecarivera/rlw-hf/hub/models--Qwen--Qwen3-14B/snapshots/40c069824f4251a91eefaf281ebe4c544efd3e18",
|
| 3 |
+
"data": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/qwen/arms/TASKW-rand-qwen.jsonl",
|
| 4 |
+
"run": "/workspace-vast/jfonsecarivera/git/bad-rl/flo/task_welfare/qwen/runs/rand-s0",
|
| 5 |
+
"steps": 0,
|
| 6 |
+
"batch": 16,
|
| 7 |
+
"micro": 1,
|
| 8 |
+
"seq_len": 4096,
|
| 9 |
+
"lr": 0.0001,
|
| 10 |
+
"warmup": 10,
|
| 11 |
+
"lora_rank": 64,
|
| 12 |
+
"lora_targets": "all-linear",
|
| 13 |
+
"lora_targets_file": "",
|
| 14 |
+
"seed": 0,
|
| 15 |
+
"bos": false,
|
| 16 |
+
"no_merge": false,
|
| 17 |
+
"save_at": "",
|
| 18 |
+
"replay": "",
|
| 19 |
+
"replay_frac": 0.0,
|
| 20 |
+
"constant_lr": false,
|
| 21 |
+
"schedule": "hf-cosine",
|
| 22 |
+
"min_lr_frac": 0.0,
|
| 23 |
+
"warmup_ratio": 0.05,
|
| 24 |
+
"lora_dropout": 0.05,
|
| 25 |
+
"no_pack": true,
|
| 26 |
+
"chat_sft": true,
|
| 27 |
+
"data_frac": 1.0,
|
| 28 |
+
"base_adapter": "",
|
| 29 |
+
"ckpt_every": 25,
|
| 30 |
+
"kill_after": 0,
|
| 31 |
+
"train_token_rows": "",
|
| 32 |
+
"method": "doc_lm_lora",
|
| 33 |
+
"world": 1
|
| 34 |
+
}
|