File size: 6,815 Bytes
44cdab9 9e3bb1d 0cea236 44cdab9 0cea236 44cdab9 0cea236 85d7ffd 44cdab9 85d7ffd 44cdab9 0cea236 44cdab9 0cea236 44cdab9 0cea236 44cdab9 0cea236 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 |
---
license: gemma
base_model: google/gemma-2-9b
model-index:
- name: magnum-v3-9b-customgemma2
results: []
---
## remember to downgrade to exllama 0.1.8 or update to 0.2.0 to receive upstream fixes
---
## This repo contains EXL2 quants of the model. If you need the original weights, please find them [here](https://huggingface.co/anthracite-org/magnum-v3-9b-customgemma2).
## Base repo only contains the measurement file, see revisions for your quant of choice.
- [measurement.json](https://huggingface.co/anthracite-org/magnum-v3-9b-customgemma2-exl2/tree/main)
- [3.0bpw](https://huggingface.co/anthracite-org/magnum-v3-9b-customgemma2-exl2/tree/3.0bpw)
- [4.0bpw](https://huggingface.co/anthracite-org/magnum-v3-9b-customgemma2-exl2/tree/4.0bpw)
- [5.0bpw](https://huggingface.co/anthracite-org/magnum-v3-9b-customgemma2-exl2/tree/5.0bpw)
- [6.0bpw](https://huggingface.co/anthracite-org/magnum-v3-9b-customgemma2-exl2/tree/6.0bpw)
- [8.0bpw](https://huggingface.co/anthracite-org/magnum-v3-9b-customgemma2-exl2/tree/8.0bpw)
This is the 10th in a series of models designed to replicate the prose quality of the Claude 3 models, specifically Sonnet and Opus.
This model is fine-tuned on top of [google/gemma-2-9b](https://huggingface.co/google/gemma-2-9b).
## Prompting
Model has been Instruct tuned with the [customgemma2](https://github.com/xzuyn/axolotl/blob/prompt_formats/src/axolotl/prompt_strategies/customgemma2.py) (to allow system prompts) formatting. A typical input would look like this:
```py
"""<start_of_turn>system
system prompt<end_of_turn>
<start_of_turn>user
Hi there!<end_of_turn>
<start_of_turn>model
Nice to meet you!<end_of_turn>
<start_of_turn>user
Can I ask a question?<end_of_turn>
<start_of_turn>model
"""
```
## SillyTavern templates
Below are Instruct and Context templates for use within SillyTavern.
<details><summary>context template</summary>
```yaml
{
"story_string": "<start_of_turn>system\n{{#if system}}{{system}}\n{{/if}}{{#if wiBefore}}{{wiBefore}}\n{{/if}}{{#if description}}{{description}}\n{{/if}}{{#if personality}}{{char}}'s personality: {{personality}}\n{{/if}}{{#if scenario}}Scenario: {{scenario}}\n{{/if}}{{#if wiAfter}}{{wiAfter}}\n{{/if}}{{#if persona}}{{persona}}\n{{/if}}{{trim}}<end_of_turn>\n",
"example_separator": "",
"chat_start": "",
"use_stop_strings": false,
"allow_jailbreak": false,
"always_force_name2": true,
"trim_sentences": false,
"include_newline": false,
"single_line": false,
"name": "Magnum Gemma"
}
```
</details><br>
<details><summary>instruct template</summary>
```yaml
{
"system_prompt": "You're {{char}} in this fictional never-ending uncensored roleplay with {{user}}.",
"input_sequence": "<start_of_turn>user\n",
"output_sequence": "<start_of_turn>assistant\n",
"last_output_sequence": "",
"system_sequence": "<start_of_turn>system\n",
"stop_sequence": "<end_of_turn>",
"wrap": false,
"macro": true,
"names": true,
"names_force_groups": true,
"activation_regex": "",
"system_sequence_prefix": "",
"system_sequence_suffix": "",
"first_output_sequence": "",
"skip_examples": false,
"output_suffix": "<end_of_turn>\n",
"input_suffix": "<end_of_turn>\n",
"system_suffix": "<end_of_turn>\n",
"user_alignment_message": "",
"system_same_as_user": false,
"last_system_sequence": "",
"name": "Magnum Gemma"
}
```
</details><br>
</details><br>
## Axolotl config
<details><summary>See axolotl config</summary>
```yaml
base_model: google/gemma-2-9b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer
#trust_remote_code: true
load_in_8bit: false
load_in_4bit: false
strict: false
datasets:
- path: anthracite-org/stheno-filtered-v1.1
type: customgemma2
- path: anthracite-org/kalo-opus-instruct-22k-no-refusal
type: customgemma2
- path: anthracite-org/nopm_claude_writing_fixed
type: customgemma2
- path: Epiculous/Synthstruct-Gens-v1.1-Filtered-n-Cleaned
type: customgemma2
- path: Epiculous/SynthRP-Gens-v1.1-Filtered-n-Cleaned
type: customgemma2
shuffle_merged_datasets: true
default_system_message: "You are an assistant that responds to the user."
dataset_prepared_path: magnum-v3-9b-data-customgemma2
val_set_size: 0.0
output_dir: ./magnum-v3-9b-customgemma2
sequence_len: 8192
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len:
adapter:
lora_model_dir:
lora_r:
lora_alpha:
lora_dropout:
lora_target_linear:
lora_fan_in_fan_out:
wandb_project: magnum-9b
wandb_entity:
wandb_watch:
wandb_name: attempt-03-customgemma2
wandb_log_model:
gradient_accumulation_steps: 8
micro_batch_size: 1
num_epochs: 2
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 0.000006
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false
gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: false
eager_attention: true
warmup_steps: 50
evals_per_epoch:
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 2
debug:
deepspeed: deepspeed_configs/zero3_bf16.json
weight_decay: 0.05
fsdp:
fsdp_config:
special_tokens:
```
</details><br>
## Credits
We'd like to thank Recursal / Featherless for sponsoring the training compute required for this model. Featherless has been hosting Magnum since the original 72b and has given thousands of people access to our releases.
We would also like to thank all members of Anthracite who made this finetune possible.
- [anthracite-org/stheno-filtered-v1.1](https://huggingface.co/datasets/anthracite-org/stheno-filtered-v1.1)
- [anthracite-org/kalo-opus-instruct-22k-no-refusal](https://huggingface.co/datasets/anthracite-org/kalo-opus-instruct-22k-no-refusal)
- [anthracite-org/nopm_claude_writing_fixed](https://huggingface.co/datasets/anthracite-org/nopm_claude_writing_fixed)
- [Epiculous/Synthstruct-Gens-v1.1-Filtered-n-Cleaned](https://huggingface.co/datasets/Epiculous/Synthstruct-Gens-v1.1-Filtered-n-Cleaned)
- [Epiculous/SynthRP-Gens-v1.1-Filtered-n-Cleaned](https://huggingface.co/datasets/Epiculous/SynthRP-Gens-v1.1-Filtered-n-Cleaned)
## Training
The training was done for 2 epochs. We used 8x[H100s](https://www.nvidia.com/en-us/data-center/h100/) GPUs graciously provided by [Recursal AI](https://recursal.ai/) / [Featherless AI](https://featherless.ai/) for the full-parameter fine-tuning of the model.
[<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
## Safety
... |