Update 1st layer + Counter size

README.md

@@ -50,17 +50,15 @@ In particular, Bounter implements three different algorithms under the hood, dep
   ```python
   from bounter import bounter
 
-  counts = bounter(need_iteration=False)
+  counts = bounter(need_counts=False)
   counts.update(['a', 'b', 'c', 'a', 'b'])
 
   print(counts.cardinality())  # cardinality estimation
   3
   print(counts.total())  # counts accumulated across all items
   5
   ```
-
-  FIXME We do not support HLL *only* at the moment! Could be "emulated" with HashTable with minimum number of buckets.
-
+
   This is the simplest use case and needs the least amount of memory, by using the [HyperLogLog algorithm](http://algo.inria.fr/flajolet/Publications/FlFuGaMe07.pdf) (built on top of Joshua Andersen's [HLL](https://github.com/ascv/HyperLogLog) code).
 
 2. **Item frequencies: "How many times did this item appear?"**
@@ -76,7 +74,7 @@ In particular, Bounter implements three different algorithms under the hood, dep
   2
   ```
 
-  This uses the [Count-min Sketch algorithm](https://en.wikipedia.org/wiki/Count%E2%80%93min_sketch) to estimate item counts efficiently, in a **fixed amount of memory**. See the [FIXME API docs](https://github.com/RaRe-Technologies/bounter/blob/master/bounter/bounter.py) for full details and parameters.
+  This uses the [Count-min Sketch algorithm](https://en.wikipedia.org/wiki/Count%E2%80%93min_sketch) to estimate item counts efficiently, in a **fixed amount of memory**. See the [API docs](https://github.com/RaRe-Technologies/bounter/blob/master/bounter/bounter.py) for full details and parameters.
 
 As a further optimization, Count-min Sketch optionally support a [logarithmic probabilistic counter](https://en.wikipedia.org/wiki/Approximate_counting_algorithm):
 
@@ -108,7 +106,7 @@ Such memory vs. accuracy tradeoffs are sometimes desirable in NLP, where being a
 
 ----
 
-For more details, see the [FIXME API docstrings](https://github.com/RaRe-Technologies/bounter/blob/master/bounter/bounter.py).
+For more details, see the [API docstrings](https://github.com/RaRe-Technologies/bounter/blob/master/bounter/bounter.py).
 
 ## Example on the English Wikipedia
 
@@ -125,15 +123,15 @@ print(counter[u'czech republic'])
 42099
 ```
 
-The Wikipedia dataset contained 7,661,318 distinct words across 1,860,927,726 total words, and 179,413,989 distinct bigrams across 1,857,420,106 total bigrams. Storing them in a naive built-in `dict` would consume more than 17.2 GB RAM (FIXME suspiciously low number -- is this true?).
+The Wikipedia dataset contained 7,661,318 distinct words across 1,860,927,726 total words, and 179,413,989 distinct bigrams across 1,857,420,106 total bigrams. Storing them in a naive built-in `dict` would consume over 31 GB RAM.
 
 To test the accuracy of Bounter, we automatically extracted [collocations](https://en.wikipedia.org/wiki/Collocation) (common multi-word expressions, such as "New York", "network license", "Supreme Court" or "elementary school") from these bigram counts.
 
 We compared the set of collocations extracted from Counter (exact counts, needs lots of memory) vs Bounter (approximate counts, bounded memory) and present the precision and recall here:
 
 | Algorithm                         | Time to build | Memory  | Precision | Recall | F1 score
 |-----------------------------------|--------------:|--------:|----------:|-------:|---------:|
-| `Counter` (built-in)              |         FIXME | 17.2 GB |      100% |   100% |     100% |
+| `Counter` (built-in)              |     FIXME 32m  | 31 GB |      100% |   100% |     100% |
 | `bounter(size_mb=128, need_iteration=False, log_counting=8)` |         18m 08s |   128 MB | 95.02% | 97.10% | 96.04% |
 | `bounter(size_mb=1024)`           |       13m 26s |    1 GB |     100% |  99.27% |   99.64% |
 | `bounter(size_mb=1024, need_iteration=False)` |     18m 38s |   1 GB |    0.9964% | 100% | 99.82% |
@@ -143,9 +141,9 @@ We compared the set of collocations extracted from Counter (exact counts, needs
 | `bounter(size_mb=4096, need_iteration=False)` |        21m 02s  |   4 GB|    100% | 100% | 100% |
 | `bounter(size_mb=4096, need_iteration=False, log_counting=1024)` |        21m 34s |   4 GB |    100% | 99.64% | 99.82% |
 
-Bounter achieves 100% F1 score at 17x less memory, compared to a built-in `Counter` or `dict`. It is also FIXME % faster.
+Bounter achieves 100% F1 score at 31x less memory, compared to a built-in `Counter` or `dict`. It is also FIXME % faster.
 
-Even with just 128 MB (137x less memory), its F1 score is still 96.04! 
+Even with just 128 MB (250x less memory), its F1 score is still 96.04! 
 
 # Support