GaetanMichelet
commited on
Commit
•
4512ada
1
Parent(s):
f69bf6e
End of training
Browse files- README.md +7 -2
- all_results.json +5 -0
- config.json +51 -0
- eval_results.json +8 -0
- runs/Aug27_15-33-37_9mqd6r3-l/events.out.tfevents.1724768879.9mqd6r3-l.3784243.1 +3 -0
README.md
CHANGED
@@ -1,8 +1,13 @@
|
|
1 |
---
|
2 |
base_model: google/gemma-2-2b-it
|
|
|
|
|
|
|
|
|
3 |
library_name: peft
|
4 |
license: gemma
|
5 |
tags:
|
|
|
6 |
- trl
|
7 |
- sft
|
8 |
- generated_from_trainer
|
@@ -16,9 +21,9 @@ should probably proofread and complete it, then remove this comment. -->
|
|
16 |
|
17 |
# Gemma-2-2B_task-2_180-samples_config-2_full
|
18 |
|
19 |
-
This model is a fine-tuned version of [google/gemma-2-2b-it](https://huggingface.co/google/gemma-2-2b-it) on the
|
20 |
It achieves the following results on the evaluation set:
|
21 |
-
- Loss:
|
22 |
|
23 |
## Model description
|
24 |
|
|
|
1 |
---
|
2 |
base_model: google/gemma-2-2b-it
|
3 |
+
datasets:
|
4 |
+
- GaetanMichelet/chat-60_ft_task-2
|
5 |
+
- GaetanMichelet/chat-120_ft_task-2
|
6 |
+
- GaetanMichelet/chat-180_ft_task-2
|
7 |
library_name: peft
|
8 |
license: gemma
|
9 |
tags:
|
10 |
+
- alignment-handbook
|
11 |
- trl
|
12 |
- sft
|
13 |
- generated_from_trainer
|
|
|
21 |
|
22 |
# Gemma-2-2B_task-2_180-samples_config-2_full
|
23 |
|
24 |
+
This model is a fine-tuned version of [google/gemma-2-2b-it](https://huggingface.co/google/gemma-2-2b-it) on the GaetanMichelet/chat-60_ft_task-2, the GaetanMichelet/chat-120_ft_task-2 and the GaetanMichelet/chat-180_ft_task-2 datasets.
|
25 |
It achieves the following results on the evaluation set:
|
26 |
+
- Loss: 0.8710
|
27 |
|
28 |
## Model description
|
29 |
|
all_results.json
CHANGED
@@ -1,5 +1,10 @@
|
|
1 |
{
|
2 |
"epoch": 16.941176470588236,
|
|
|
|
|
|
|
|
|
|
|
3 |
"total_flos": 7.574414567879475e+16,
|
4 |
"train_loss": 0.8188091479241848,
|
5 |
"train_runtime": 3221.0584,
|
|
|
1 |
{
|
2 |
"epoch": 16.941176470588236,
|
3 |
+
"eval_loss": 0.8710008263587952,
|
4 |
+
"eval_runtime": 13.9215,
|
5 |
+
"eval_samples": 36,
|
6 |
+
"eval_samples_per_second": 2.586,
|
7 |
+
"eval_steps_per_second": 2.586,
|
8 |
"total_flos": 7.574414567879475e+16,
|
9 |
"train_loss": 0.8188091479241848,
|
10 |
"train_runtime": 3221.0584,
|
config.json
ADDED
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "google/gemma-2-2b-it",
|
3 |
+
"architectures": [
|
4 |
+
"Gemma2ForCausalLM"
|
5 |
+
],
|
6 |
+
"attention_bias": false,
|
7 |
+
"attention_dropout": 0.0,
|
8 |
+
"attn_logit_softcapping": 50.0,
|
9 |
+
"bos_token_id": 2,
|
10 |
+
"cache_implementation": "hybrid",
|
11 |
+
"eos_token_id": [
|
12 |
+
1,
|
13 |
+
107
|
14 |
+
],
|
15 |
+
"final_logit_softcapping": 30.0,
|
16 |
+
"head_dim": 256,
|
17 |
+
"hidden_act": "gelu_pytorch_tanh",
|
18 |
+
"hidden_activation": "gelu_pytorch_tanh",
|
19 |
+
"hidden_size": 2304,
|
20 |
+
"initializer_range": 0.02,
|
21 |
+
"intermediate_size": 9216,
|
22 |
+
"max_position_embeddings": 8192,
|
23 |
+
"model_type": "gemma2",
|
24 |
+
"num_attention_heads": 8,
|
25 |
+
"num_hidden_layers": 26,
|
26 |
+
"num_key_value_heads": 4,
|
27 |
+
"pad_token_id": 0,
|
28 |
+
"quantization_config": {
|
29 |
+
"_load_in_4bit": true,
|
30 |
+
"_load_in_8bit": false,
|
31 |
+
"bnb_4bit_compute_dtype": "bfloat16",
|
32 |
+
"bnb_4bit_quant_storage": "uint8",
|
33 |
+
"bnb_4bit_quant_type": "nf4",
|
34 |
+
"bnb_4bit_use_double_quant": false,
|
35 |
+
"llm_int8_enable_fp32_cpu_offload": false,
|
36 |
+
"llm_int8_has_fp16_weight": false,
|
37 |
+
"llm_int8_skip_modules": null,
|
38 |
+
"llm_int8_threshold": 6.0,
|
39 |
+
"load_in_4bit": true,
|
40 |
+
"load_in_8bit": false,
|
41 |
+
"quant_method": "bitsandbytes"
|
42 |
+
},
|
43 |
+
"query_pre_attn_scalar": 256,
|
44 |
+
"rms_norm_eps": 1e-06,
|
45 |
+
"rope_theta": 10000.0,
|
46 |
+
"sliding_window": 4096,
|
47 |
+
"torch_dtype": "bfloat16",
|
48 |
+
"transformers_version": "4.44.0",
|
49 |
+
"use_cache": true,
|
50 |
+
"vocab_size": 256000
|
51 |
+
}
|
eval_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 16.941176470588236,
|
3 |
+
"eval_loss": 0.8710008263587952,
|
4 |
+
"eval_runtime": 13.9215,
|
5 |
+
"eval_samples": 36,
|
6 |
+
"eval_samples_per_second": 2.586,
|
7 |
+
"eval_steps_per_second": 2.586
|
8 |
+
}
|
runs/Aug27_15-33-37_9mqd6r3-l/events.out.tfevents.1724768879.9mqd6r3-l.3784243.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c19a84500d9c8099666328c73bbfd6b951461646425ef334d67a134dd0d910e4
|
3 |
+
size 359
|