Hugging Face's logo

merve
/
gemma-3n-finevideo

Transformers
TensorBoard
Safetensors
Generated from Trainer
trl
sft
Model card Files Metrics Community
1
merve HF Staff commited on
Commit
4923b9d
·
verified ·
1 Parent(s): 133060b

Training in progress, step 100

Browse files
Files changed (10) hide show
  1. README.md +2 -2
  2. adapter_config.json +20 -20
  3. runs/Jun30_15-41-50_66bfc5776606/events.out.tfevents.1751298118.66bfc5776606.1352.0 +3 -0
  4. runs/Jun30_15-41-50_66bfc5776606/events.out.tfevents.1751298272.66bfc5776606.1352.1 +3 -0
  5. runs/Jun30_15-41-50_66bfc5776606/events.out.tfevents.1751298547.66bfc5776606.1352.2 +3 -0
  6. runs/Jun30_15-41-50_66bfc5776606/events.out.tfevents.1751298624.66bfc5776606.1352.3 +3 -0
  7. runs/Jun30_15-51-14_66bfc5776606/events.out.tfevents.1751298679.66bfc5776606.1352.4 +3 -0
  8. runs/Jun30_15-51-36_66bfc5776606/events.out.tfevents.1751298699.66bfc5776606.1352.5 +3 -0
  9. runs/Jun30_16-20-23_66bfc5776606/events.out.tfevents.1751300433.66bfc5776606.20495.0 +3 -0
  10. training_args.bin +2 -2
README.md CHANGED
@@ -4,8 +4,8 @@ library_name: transformers
4
  model_name: gemma-3n-finevideo
5
  tags:
6
  - generated_from_trainer
7
- - sft
8
  - trl
 
9
  licence: license
10
  ---
11
 
@@ -38,7 +38,7 @@ This model was trained with SFT.
38
  - Transformers: 4.53.0
39
  - Pytorch: 2.6.0+cu124
40
  - Datasets: 3.6.0
41
- - Tokenizers: 0.21.1
42
 
43
  ## Citations
44
 
 
4
  model_name: gemma-3n-finevideo
5
  tags:
6
  - generated_from_trainer
 
7
  - trl
8
+ - sft
9
  licence: license
10
  ---
11
 
 
38
  - Transformers: 4.53.0
39
  - Pytorch: 2.6.0+cu124
40
  - Datasets: 3.6.0
41
+ - Tokenizers: 0.21.2
42
 
43
  ## Citations
44
 
adapter_config.json CHANGED
@@ -25,33 +25,33 @@
25
  "revision": null,
26
  "target_modules": [
27
  "linear_start",
28
- "ffw_layer_2",
29
- "altup_projections.1",
30
- "post",
31
- "v_proj",
32
- "o_proj",
33
- "modality_router",
34
  "per_layer_projection",
35
- "per_layer_input_gate",
36
  "linear_left",
37
- "q_proj",
38
- "linear_end",
39
- "down_proj",
40
- "prediction_coefs",
41
- "altup_unembed_projections.0",
42
- "input_proj_linear",
43
  "k_proj",
44
- "linear_right",
45
- "altup_projections.2",
46
- "up_proj",
47
- "gate_proj",
48
- "embedding_projection",
49
  "correction_coefs",
50
- "altup_projections.0",
51
- "pos_proj",
52
  "per_layer_model_projection",
53
  "ffw_layer_1",
 
 
 
 
 
54
  "altup_unembed_projections.2",
 
 
 
 
 
 
 
 
 
 
 
55
  "altup_unembed_projections.1"
56
  ],
57
  "task_type": "CAUSAL_LM",
 
25
  "revision": null,
26
  "target_modules": [
27
  "linear_start",
28
+ "embedding_projection",
29
+ "down_proj",
 
 
 
 
30
  "per_layer_projection",
31
+ "o_proj",
32
  "linear_left",
 
 
 
 
 
 
33
  "k_proj",
 
 
 
 
 
34
  "correction_coefs",
35
+ "post",
 
36
  "per_layer_model_projection",
37
  "ffw_layer_1",
38
+ "altup_unembed_projections.0",
39
+ "altup_projections.2",
40
+ "prediction_coefs",
41
+ "ffw_layer_2",
42
+ "up_proj",
43
  "altup_unembed_projections.2",
44
+ "altup_projections.0",
45
+ "gate_proj",
46
+ "input_proj_linear",
47
+ "pos_proj",
48
+ "modality_router",
49
+ "per_layer_input_gate",
50
+ "linear_end",
51
+ "linear_right",
52
+ "altup_projections.1",
53
+ "v_proj",
54
+ "q_proj",
55
  "altup_unembed_projections.1"
56
  ],
57
  "task_type": "CAUSAL_LM",
runs/Jun30_15-41-50_66bfc5776606/events.out.tfevents.1751298118.66bfc5776606.1352.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad4acb284864b4e9b88980881d6beb3895a08e42bbe1ee7d7293d1bc79895ae4
3
+ size 9273
runs/Jun30_15-41-50_66bfc5776606/events.out.tfevents.1751298272.66bfc5776606.1352.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:87c1dcb8aa8da47e6e425a8aea93203307a31c9ac53be6a5b4295df1bc6e459a
3
+ size 10683
runs/Jun30_15-41-50_66bfc5776606/events.out.tfevents.1751298547.66bfc5776606.1352.2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5aac8a121067bede967a6cfabccee971d2f3837438284639289e73072ca30ead
3
+ size 9273
runs/Jun30_15-41-50_66bfc5776606/events.out.tfevents.1751298624.66bfc5776606.1352.3 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f511e5f0ef559d2cc1a0018d7406dadefd275e4412b52dd302970a6ed0f6fc9
3
+ size 9273
runs/Jun30_15-51-14_66bfc5776606/events.out.tfevents.1751298679.66bfc5776606.1352.4 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:46aeba50ff979a56e5a7e62710414a371a9eb6f2bba21c508abfccff2dfd157c
3
+ size 9275
runs/Jun30_15-51-36_66bfc5776606/events.out.tfevents.1751298699.66bfc5776606.1352.5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ff62f7f6816e725b1a2d2016ced98f5fd21e68d5f6e69c663e529bedb72cb154
3
+ size 9599
runs/Jun30_16-20-23_66bfc5776606/events.out.tfevents.1751300433.66bfc5776606.20495.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a6d13fee2c776205c33e1cd8d20c02086f47d2279a8f4cdbd2fe7b99bf53e78
3
+ size 21702
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:19ce49882531a204abe2db6d6d72e853d245395005a716b5499c2953a201b438
3
- size 5752
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f5520e030c7191bac4d3099929fe303545ea8469ce88179ab152aa2d7c7dede3
3
+ size 5816